Eyebrows for a Heart-Shaped Face: Peak Position, Brow Height and the Forehead

What is the best eyebrow shape for a heart-shaped face?

Five. That is how many statements in the whole record tie a brow to this face shape by name, across twelve years and eleven months. Thirteen dated exchanges contain both the shape and a sentence about the eyebrows. Four of the thirteen name the shape as the reason for what they say about the brow; the fifth statement that does so is the clinical one, and it is not an exchange at all. In the other nine the brow sentence is generic or the shape is mentioned and then dropped. The volume of talk about this outline is large. The volume of talk about its brows is not.

Split the five by who was speaking and the number gets smaller. Two are advice given to another person. September 2013: the reply with the most support in its exchange says the classic brow shape belongs to an oval face, and that applying it to a heart face was the error — a diagnosis, not a prescription, and it never says what to do instead. December 2024: sharp accentuates a heart-shaped face, and reads as more sophisticated and mature than the alternatives on offer. Two are people describing their own faces rather than advising anyone. December 2013: a very heart-shaped face whose brows are naturally angled and will not go into a round shape. April 2026: medium-thin arched brows that suit a heart face with high cheekbones and a strong jawline. The fifth is clinical, and it is the only one of the five that names a landmark.

Read the two pieces of actual advice as instructions and they point in opposite directions. Sharp, from 2024. Round, from a peer-reviewed review in 2026. The word is not being used loosely on either side: the same review's round-face row forbids round brows to that face on the grounds that they turn it into a beach ball, and separately says its peak is best moved out toward the end of the brow. So round is a shape that this table prescribes here and prohibits there, and the record's one positive recommendation for this face is its opposite.

The third position is the one nobody wrote down. In June 2026 someone with an oval-or-heart face, naturally straight and barely arched brows, and thicker darker hair in them than most, reported that the combination reads as masculinising on her, noted that it does not go for everyone, and asked what the reasoning was behind people with her shape being advised against rounded and arched brows. The exchange drew one reply, and that reply is automated. The question was never answered. The reasoning she asked for does not exist anywhere in the record — not on the round side, not on the sharp side, not in the review published eighteen days later, which reaches the opposite conclusion by a different route and never mentions masculinisation.

So the direct answer to the heading is that there is no consensus to report, and that the absence carries more information than a consensus would. Two of the five statements are self-reports about the writer's own face. One is a clinical prescription whose table is reproduced from another publication and contains no measurement. One says only that the standard shape belongs to a different face. That leaves a single word of positive advice for this outline in twelve years and eleven months, and it is contradicted by the only source in the record that states a method.

What the five agree on is narrower than it looks, and the narrowing is the finding. One of them excludes a shape without naming a replacement. Three of the remaining four are adjectives about the bend — sharp, naturally angled, arched — and the fourth is the only one that can be checked: a round brow whose peak sits between two landmarks you can find on your own face. So the agreement is not about a shape. It is that the bend is the thing worth arguing about, and four times out of five the argument is conducted in adjectives that no source defines. Not one of the five says anything about the other setting — how much forehead the brow leaves above it — which is why section five has to build that one out of measurements rather than out of advice.

Which face is this page for — heart, inverted triangle, or an oval leaning heart?

The record cannot agree, and the disagreement is about which feature defines the shape. The most-offered discriminator is a widow's peak, which is a hairline feature and sits above the brow rather than in it. In one April 2023 exchange that ran to hundreds of replies, the person asking was given heart, inverted triangle, diamond, square, oval and three combinations. One reply states the rule plainly: a heart-shaped face differs from an inverted triangle only in that a heart has a widow's peak, so wouldn't she technically be inverted triangle — followed, in the next sentence, by an admission that the writer did not know how that would affect contouring. Another says the chin is not sharp enough for heart and the cheeks are too sculpted. Another says square, because the forehead lacks the heart outline and there is no widow's peak. The person asking then confirmed she does have one, hidden by how her hair was pinned. The classification did not settle.

The clinical definition uses different features entirely. The 2026 review's table gives this outline as a forehead and cheekbones of the same width, a narrower jaw, a more angled chin, long straight jaw lines, and length greater than width. No hairline. No brow. Two definitions of one shape, keyed to opposite ends of the face, and the thing the whole page is about is not part of either.

That table also has a hole worth naming, because it decides who this page is for. The sentence introducing it says there are six primary facial shapes and then lists five: oval, round, square, heart and long. The table itself has five rows. Neither the sentence nor the table contains diamond, and neither contains inverted triangle — the two names the record argues about most. A 2026 clinical source that prescribes brows by face shape does not recognise either of the categories that most of the people asking were placed in.

This site classifies seven shapes and keeps inverted triangle inside heart, and says so on its about page rather than quietly absorbing it. The measurement that decides is the forehead ratio: forehead width over cheekbone width. On the site's own constants this outline scores 0.978 there, the highest of the seven, meaning the forehead is 97.8 per cent as wide as the cheekbones — the only shape on the site whose forehead approaches the widest point of the face. Diamond, its nearest named boundary, scores 0.91, and the two are separated by that one feature at 3.09 standard deviations and by nothing else: their jaw ratios are identical at 0.759 and their length ratios differ by 0.27 of a standard deviation. One caveat the site publishes and this page repeats rather than hides: the heart prototype was interpolated along the literature's ordering from a single calibrated example, not measured across a group, so those separations carry that weight and no more.

The jaw, meanwhile, is the feature doing the least work. At 134.9 degrees this outline's jaw angle sits at the median of the seven, which is why the clinical table's insistence on a narrower jaw and a more angled chin separates it from things this site's numbers do not. Two classification systems, both defensible, choosing different features — and the record picked a third, the hairline.

One more case, and it is the most recent thing in the record. In August 2026 a person who had always assumed she was oval wrote that with age her chin seemed less rounded and her cheekbones a little wider, wondered about diamond, and added that her face is very long. The answer used the forehead, not the length: your forehead looks like the widest part, so heart leaning slightly oval, and a face can carry more than one shape. Length and forehead width are different axes. The answer picked the one that decides, and this site reports a second shape with a margin against each rather than forcing one, for exactly that reason.

The outline is also not fixed, which matters if you are about to remove hair on the strength of a name. In September 2020 someone reported that dropping five or six pounds changed her face from round to heart or upside-down triangle, with a more prominent jawline and cheekbones and the fat gone from her chin, jaw, neck and cheeks. Six pounds. A brow decision made against a shape name is made against something that moved once and can move back, which is a further argument for the two settings in the intro — they are keyed to landmarks on the brow itself, and landmarks do not change with weight.

Where the peak goes: four sources, one place, none of them citing another

The textbook answer is a vertical line. A February 2016 review of ideal brow shape and position — published online in October 2014 — puts the apex on a vertical directly above the lateral limbus, with the medial and lateral ends of the brow at approximately the same horizontal level, the medial brow beginning on the same vertical plane as the lateral extent of the ala and the inner canthus, and the lateral end finishing on an oblique line drawn from the most lateral point of the ala through the lateral canthus. The same review warns that placing the peak too far inward creates an undesired surprised look, and that a low medial brow with a high lateral peak induces an angry one. It also says, in so many words, that it is impossible to define an ideal eyebrow suitable for every face.

The measured answer sits further out. A July 2017 observational case series photographed 60 women five times each while they positioned their own ideal brow with their fingers. Where their peaks actually were: between the lateral limbus and the lateral canthus in 58.3 per cent of them, which is 35 of 60; at the lateral canthus in 28.3 per cent, or 17; at the lateral limbus in 10 per cent, or 6; and no detectable peak at all in 3.3 per cent, or 2. Those four categories account for all 60. Where they wanted the peak: at the lateral canthus for 33.3 per cent, and between the canthus and the limbus for 66.6 per cent. Those two account for 99.9 per cent, which leaves 0.1 per cent for any other position — and the position the textbook prescribes is the one that 10 of the 60 actually had.

The clinical answer for this face is between, not at. The 2026 review's heart row specifies a round brow shape with the peak in between the lateral limbus and the lateral canthus line, as the thing that softens and balances the face. That is the only peak position anywhere in the record prescribed to this outline by name, and it is the position 58.3 per cent of the 60 already had.

The answer given to a person is a fraction, and it is the same fraction. October 2024, to a person who had shaved her brows off by accident and was choosing between options: your arch needs to go about two-thirds toward your temple and not halfway on your eyebrow. A second reply in the same exchange, written the same day: some of these have your arch matching up with your pupil and it looks very strange, your arch should be past the pupil of your eye, not dead centre with it. Neither was answering a question about this face shape. Neither had read either paper.

There is a fifth, and it arrived inside one of the four. The same 2026 review, in its section on gender differences rather than its face-shape table, states that the feminine brow peaks at the junction between the middle and lateral third of the brow. The junction between the middle and lateral third is two-thirds. Same place, different reason, same document — which is why this page counts four independent sources and not five.

Converted, so that the fraction stops being a mood. Halfway along a brow of length L is L/2; two-thirds along is 2L/3; the gap is L/6. On a 50 mm brow that is 8.33 mm, or 0.33 in. On 55 mm, 9.17 mm and 0.36 in. On 60 mm, 10.00 mm and 0.39 in. On 65 mm, 10.83 mm and 0.43 in. Every inch figure is the millimetre figure divided by 25.4. The whole disagreement between 'give it an arch' and 'give it an arch, but out near the temple' is a sixth of your brow length, and it is roughly a centimetre for most people.

The caveat, stated before anyone has to ask for it. None of those 60 women was classified by face shape — the words face shape do not appear in that study at all. Its authors attribute the differences between their landmarks and other studies' to variation in race, gender, culture and age, and note that there is no uniform rule for planning brow surgery. So what the four-source convergence establishes is that a position is stable across a textbook, a measurement, a clinical table and lay advice. It does not establish that the position is right for this outline. Only one of the four was addressing this outline, and it did so in a table reproduced from someone else's publication.

The middle third of a brow runs from its inner start to one-third of its length; the lateral third from two-thirds to the end. The junction between them is therefore at two-thirds of the total, which is why a 2026 clinical statement about gender and a 2024 answer about temples name the same place. The 58.3%, 28.3%, 10% and 3.3% are 35, 17, 6 and 2 of 60; 33.3% and 66.6% are the same study's desired-position figures and sum to 99.9%. Landmarks: the lateral canthus is the outer corner of the eye, the lateral limbus is the outer edge of the iris. Sources are described here by kind and date and are never linked.

135 degrees, and the corner that is actually named

One number in the entire record states which corner it is measuring. The July 2017 study defines the brow apex angle as the angle between the medial portion of the brow — usually, it says, the medial two-thirds — and the tail, the lateral one-third. The mean desired apex angle across the 60 was 135 degrees with a standard deviation of 9. The paper prints the unit as millimetres, which is a typographical error; its own second table gives 134.66 ± 8.12 and 135.44 ± 9.22 for the two age groups, and those are degrees.

That definition is more useful than the number, and it is also doing something worth catching. The angle is measured between a segment defined as two-thirds of the brow and one defined as a third — so the position claim from the previous section is built into the angle's own definition. The two-thirds peak and the 135-degree bend are not two independent pieces of evidence. They are one study, stating the same thing twice, and the lay answer in the table above is the only source that reaches two-thirds from somewhere else.

What 135 degrees does, geometrically. Treat the brow as two straight segments meeting at the peak: a medial run of 2L/3 and a tail of L/3, with interior angle A at the join. The perpendicular distance from the tip of the tail to the medial line is then exactly (L/3) × sin(180° − A), whatever tilt the medial line itself has, because the distance is measured against that line rather than against the horizontal. Measured straight down instead, the figure is larger by a factor of 1 ÷ cos(θ), where θ is the angle the medial run rises at — under one per cent for any rise up to about eight degrees, so the two are interchangeable on a real brow and only the first is exact. On a 60 mm brow the tail is 20 mm, and 135 degrees puts the tip 14.14 mm from the medial line, which is 0.56 in. At 155 degrees it is 8.45 mm and 0.33 in. At 165, 5.18 mm and 0.20 in. At 180 — a straight brow — it is 0.00 mm, and the tail simply continues the line.

The spread matters more than the mean. A standard deviation of 9 degrees, on that same 60 mm brow, runs from 11.76 mm at 144 degrees to 16.18 mm at 126: a band 4.42 mm wide, or 0.17 in, covering one standard deviation on each side of the mean. Ten degrees of apex angle near the middle of the range is worth 2.47 mm. So the number is not a target you can hit; it is a band about four and a half millimetres across. Read what the band does and does not contain, because that is the useful part. It contains the whole distance between arched and sharper-than-arched, which is 2.04 mm. It does not contain a straight brow, which sits at 180 degrees, five standard deviations from the mean, with no offset at all — and a straight brow against an arched one is the argument this page opened with. The measured spread describes how much people vary around one kind of brow. It says nothing about the distance between kinds.

The part usually left out of the citation is the preference split. Of the 60, 48.33 per cent wanted a wider apex angle than the one they had, 28.33 per cent a narrower one, and 23.33 per cent no change. The modal preference is to open the bend, and it is not a majority: 29 of 60 wanted wider, 17 narrower, 14 left alone. Age made no statistical difference. The over-forty group, 15 of them, averaged 134.66 degrees with a standard deviation of 8.12; the forty-and-under group, 45 of them, averaged 135.44 with a standard deviation of 9.22. The difference between the two means is 0.78 of a degree, which is under a tenth of the 9-degree standard deviation on either group and under a sixth of the millimetre band computed two paragraphs up.

Marked as this page's inference rather than anyone's finding: none of the 60 was classified by face shape, so nothing in this section is a heart-face number and none of it should be read as one. What the angle is good for here is narrower and still useful. It is the only defined corner in the record, which makes it the only thing that can convert an adjective into a millimetre — and the record's adjectives are what the argument is actually made of. They are not scattered, either. The 2026 review lists five brow shapes: curved, sharp-angled, soft-angled, rounded and flat. It adds that arched configurations are separately classified into low, medium and high arches, and calls the result eight primary brow characteristics. Eight names, from one list, and not one of them is given a measurement or an angle anywhere in the document. Its table attaches a landmark twice and both times to a position rather than to a shape: the peak between the lateral limbus and the lateral canthus for this face, the peak directly above the angle of the jaw for a square one. Neither says how much the brow bends. Three further names the record argues about on this face appear in no list at all — classic, fluffy, thin and straight — and one exchange uses them as captions on numbered photographs, where a reply picks 'eight, straight, and nine, sharp' and another asks for 'not thick and fluffy', with nobody defining a word. Against a stated corner the names become a band, and a band can be checked.

Brow height against the forehead sitting under it

This is the section that belongs to this face and to no other, and it follows from one measurement. Of the seven outlines this site classifies, this is the only one whose forehead measures wider relative to its cheekbones than any other's — 0.978 against the next highest of 0.951. The forehead is the widest thing about the face. The brow is its lower boundary. So the gap between the two is not a brow measurement on this outline. It is a forehead measurement taken from underneath, and every instruction that removes hair from the top of the brow is an instruction that enlarges the widest part of the face.

The measured number, with its definition, because the definition is what makes it comparable. Brow height in the July 2017 study is the vertical distance from the top of the brow to the horizontal line passing through the inferior boundary of its medial portion. Across the 60 women it averaged 7.51 mm with a standard deviation of 1.85, and ran from 4 to 13 mm. In inches: 0.30 in on average — 7.51 divided by 25.4 is 0.2957 — with a standard deviation of 0.07 in and a range of 0.16 to 0.51 in. A separate survey reached 13 mm as well, which is the top of that range and the reason the ceiling here is not an outlier.

The forehead it sits under. Average forehead height is approximately 5 cm in women and 6 cm in men: 50 mm, which is 1.97 in, against 60 mm and 2.36 in, a difference of 10 mm or 0.39 in. Set the two side by side and the brow gap is 15.02 per cent of a 50 mm forehead — 7.51 divided by 50 — and 12.52 per cent of a 60 mm one. Put differently: of the 57.51 mm between hairline and the underside of the brow on an average female face, the brow gap is 13.06 per cent. It is a small share of the upper face and a large share of itself, and both of those are true at once, which is exactly why the trade below is lopsided.

The trade, worked. Removing hair from the top edge lowers the top of the brow, which reduces brow height by the amount removed and increases forehead height by the same amount. Take 1 mm from a brow at the mean: 7.51 becomes 6.51 mm, a loss of 13.3 per cent of the brow, while the forehead goes from 50 to 51 mm, a gain of 2.0 per cent, or 0.04 in. Take 2 mm: 5.51 mm left, 26.6 per cent of the brow gone, forehead up 4.0 per cent and 0.08 in. Take 3 mm: 4.51 mm left, 39.9 per cent gone, forehead up 6.0 per cent and 0.12 in. At the mean the exchange rate is 6.7 to 1 against you — 13.32 divided by 2.0.

Your starting height sets the price, and nothing else does. On a 4 mm brow, the bottom of the measured range, that same 1 mm is 25.0 per cent of the brow for the same 2.0 per cent of forehead: 12.5 to 1. On a 13 mm brow it is 7.7 per cent for 2.0 per cent: 3.8 to 1. The forehead side never changes, because 1 mm cut is 1 mm of forehead whoever you are. The brow side is entirely a function of what you started with. That is why the same instruction is cheap for one person and ruinous for another, and why neither of them can tell which she is from the instruction alone — you have to measure first, and no source in the record asks anyone to.

The one place in the pool where the forehead is named as the reason is not about this face. A November 2024 exchange, whose subject was something else, produced this: never take hair from the top of the brows, meaning the side toward the forehead, and remove from the bottom instead, because you have forehead space and this will give a more aesthetic ratio — with the note that the opposite applies to smaller foreheads. It is the only entry anywhere in the record that states the mechanism this section has just quantified, it states it correctly, and it is not in the table below because it was not addressed to this outline. That a mechanism this load-bearing appears once, in passing, about a different face, is the single clearest measure of how little of this advice is reasoned rather than repeated.

Two numbers that are not brow height and get quoted as if they were. The same 2017 study measured a different gap — from the upper border of the lid crease to the lower edge of the brow — at three points, and got 3 to 18 mm at the mid pupil, 4 to 19 mm at the lateral limbus, and 12 to 30 mm at the lateral canthus. Elsewhere in the literature that lid gap is put at 7 to 12 mm, and 15 to 16 mm has been suggested as the ideal. All five of those figures are lid gaps. Brow height is the 7.51 mm figure, measured from the top of the brow downward to the medial underside, and the two are not convertible into each other. Any instruction that quotes a millimetre figure without saying which gap it means cannot be followed, and the record contains several.

The three starting values are the floor, the mean and the ceiling of a measured range across 60 women (mean 7.51 mm, standard deviation 1.85). Brow height is the vertical distance from the top of the brow to the horizontal line through the inferior boundary of its medial portion. The forehead side of the trade is the same in every row: a 1 mm cut adds 1 mm to a 50 mm forehead, which is 2.0% and 0.04 in; 2 mm adds 4.0% and 0.08 in; 3 mm adds 6.0% and 0.12 in. That makes the exchange rate 12.5 to 1 against on a 4 mm brow, 6.7 to 1 on the mean, and 3.8 to 1 on a 13 mm brow. Inch figures are the millimetre figure divided by 25.4, exact since 1959, rounded to two decimals.

Everything above is the reasoning. The sheet is the doing: a tracing method for settling which outline you actually have, the four brow landmarks in order from the nose outward, the tail-offset sweep from a straight brow to a 126-degree bend at four brow lengths, the forehead trade at three starting heights, the regrowth clock with the arithmetic shown, and a decoder that turns the shape names in circulation into a number: the five a clinical review lists, converted to an angle and an offset, the three further names it never defines, and the arch-height axis that measures the peak's rise instead of the bend. Printable, A4, and it has blanks to write in: Download the heart-face brow chart (printable) ↓

The top edge: forbidden five times, defended once, by a round face

Search heart shape eyebrows and the top edge is where the advice concentrates, and it concentrates as a prohibition. One exchange in October 2024, from a person who had shaved her brows off by accident and was asking about thickness, length and arch, produced five separate statements against removing hair from the top. The second-most-supported answer in it is one sentence long: the best advice I got is to never pluck the tops of your brows. Then, from four different people: do not pluck the top edge and the arch so much, and fill it in a little instead; please never pluck the tops again; stop plucking or removing from the top of your brows; do not pluck the top ever, pluck the bottom.

One person defended it, two days later, with the only mechanism anywhere in the record. Plucking the top of her brows was the best decision she ever made, she wrote, because hers are very thick vertically and it helped make them look longer horizontally — which helps, she said, because her face is round. A round face. The single reasoned defence of the instruction that this face shape is told never to follow was given by someone with a different outline, about a different problem, and the two things she names as her reason are vertical thickness and roundness.

She walked it back the next day. Asked whether she had said it would work for the asker's face, she said no, her point was that plucking the top can work for some people, and she did not want anyone to think they should never do it entirely. Which people, is not said. What the walk-back removes is the only part that was useful: the mechanism was specific — vertically thick brows, read as longer horizontally — and 'some people' is not.

Read against the arithmetic in section five, the mechanism turns out to be the trade with its signs flipped. Making a vertically thick brow look longer horizontally means paying brow height to buy brow length. On a round face that is a purchase, because the outline has width to spare and no length. On this outline the height you pay is forehead, and the forehead is already the widest thing on the face — so the same move buys nothing and costs the 6.7 to 1 that the table prices. The defence is not wrong. It is addressed to a face that gains from it, and it was then generalised, in the walk-back, to everyone.

The other side of the top edge, from people describing what happened. A July 2023 asker with a narrow heart-shaped face wrote that her thick black brows are untameable, that no glue, soap or gel will hold them, that she gets them laminated every now and then and they last a week before going back to how they were and then fall out and leave bald patches, and that the professionals do not know what to do with them. She asked whether to make them more arched or more flat, shorter or longer, thin or grown out. That exchange has no replies. It is the most precisely posed question about this face in the entire record, from the person with the most specific version of the problem, and it drew nothing.

An August 2023 asker had shaved the ends of her brows for years and was growing them back. She had always thought her face was heart-shaped, was told she is closer to square given the width from her temples to her jawline and her smaller, less pronounced chin, was told that the triangle and heart do not fit her face at all, and was sent to look up eyebrow mapping. An August 2024 asker, who named a heart face and described an inner half that is dark and mostly straight and an outer half that is curly with lighter hairs intermixed so that it looks sparse, got six answers. Not one of the six uses the shape she named.

What none of the five prohibitions contains is an amount. Not a millimetre figure, not a hair count, not a line to stop at, not a way to check whether you have already gone too far. On most of the face that would be a small omission. On this outline it is the whole problem, because the cost of the instruction is a function of the starting height, and the starting height spans 4 to 13 mm — a factor of three and a quarter, and the difference between paying 7.7 per cent of your brow and paying 25 per cent of it for the same millimetre.

How long a mistake lasts, computed

The record's standard reassurance is that it will grow back. That is true, and the duration is the part nobody supplies, so here it is, computed from published phase lengths. Eyebrow hair spends 2 to 3 months in anagen, 2 to 3 weeks in catagen and 2 to 3 months in telogen. Summing the lower bounds: 2 months, plus 2 weeks, plus 2 months. Converting the weeks at 4.345 weeks per month — 52.14 weeks in a mean year, divided by twelve — gives 0.46 months, for a total of 4.46. Summing the upper bounds the same way gives 3 + 0.69 + 3 = 6.69 months. In days, at 30.44 days per mean month: 136 to 204.

So a plucked hair on this face is a four-and-a-half to six-and-a-half month commitment at minimum, and that is before counting the time to become visible rather than merely present. The comparison that makes it concrete: scalp hair spends 2 to 8 years in anagen. The shortest scalp growth phase is eight times the longest brow growth phase. A brow mistake is not a haircut, and the record treats it as one.

What the record does say about permanence is thin and mostly second-hand. An August 2024 reply cautions the asker that there are a few strays she could have removed by a professional if she does not mind the upkeep, but to remove very little. Another reply in the same exchange reports paying a technician who then said she did not need to do very much at all, being annoyed about it at the time, and concluding it was worth going to someone who knew what would suit her face rather than ending up with generic brows — and that she took photographs afterward and now uses them as her guide. Neither states a duration. Neither states a price. The photographs are the only reversal mechanism anyone in the record offers.

Where this page will not fill a gap, and says so instead. No source in the record states how long a lamination is supposed to last, so the July 2023 report of one week cannot be compared against an expected figure — only against the hair fall that followed it, which no source dates either. Implanted pigment appears in the pool only as a service that is offered, never with a duration attached. The one clinical timepoint anywhere in the record is 30 days, in the 2026 review's heart-face case, and that is an assessment point rather than a duration: the review does not say how long the effect lasted. Three numbers that would settle real decisions, and the record does not contain them.

One thing the regrowth clock does settle, and it reorders the whole page. If an error costs four and a half to six and a half months, then the reversible moves are not merely safer, they are the only ones that can be evaluated. Setting the hairs where they already are washes off the same evening. Changing colour washes off in a similar timeframe. Taking hair off the top edge does not wash off at all, and on this outline it is also the move with the worst exchange rate in section five. That ordering — set, then colour, then outline, then removal — is not a preference. It is what the durations force.

Where the advice went instead: fringe, contour, colour and a referral

Nine of the thirteen dated exchanges name this face and then either give a brow instruction that has nothing to do with it, or give no brow instruction at all. Add the exchanges in the pool that name the outline and contain no brow sentence at all, and most of what was said to people with this face was about something else. In date order, and none of it is wrong — it is just not the brow.

The fringe and the part. October 2024: you have a heart-shaped face with a high forehead, so you need some wispy bangs, and maybe try a side part to change it up from time to time — offered as what the writer's own stylist suggested for the same combination. June 2023: curtain bangs again, in a general improvement list, alongside thicker brows and defined lips. September 2023: a person already wearing curtain bangs who felt her forehead was still a glaring issue and asked whether a fringe or a side swoop would cover it better. And the useful one, from August 2016: a high forehead described as a neutral feature, like a pointy chin or high cheekbones or a heart-shaped face, most of which are considered attractive — a way of marking what makes a face distinct rather than a list of bad ones. The same reply then says the only reason anyone would recommend bangs is that they tend to look good on people with bigger foreheads, so they could be cute if she wanted to experiment, but that she definitely does not need them. Every other fringe entry in the pool offers the fringe as a correction. This one offers it as an option and says the face is fine without it.

Blush and contour, which is where most of it went. March 2019, to a person who said twice that she thought her face was heart-shaped: smile and put the blush on the part of your cheek that sticks out most, then blend up toward your temples a little; for contour, suck your cheeks in, and the part that does not suck in is the cheekbone, with the contour going in the hollow below it and stopping where it lines up with the edge of the eye. June 2026: contour for a heart face shape, more of it to accentuate the cheekbones provided it is blended, blush toward the temples for a snatched look or on the apples toward the front for a cute one, highlight on the nose — and one judgement rather than a placement — that the face reads as lower in contrast, so softer colours that are not dull. April 2023, from someone with this face and higher-set cheekbones: she contours at an angle for a sharper look, which she did when younger, and now prefers a reverse contour with blush.

Colour, and the one exchange where colour was the entire answer. April 2022: a person with a heart face, dirty blonde hair, green eyes, neutral skin tone, naturally taupe brows, one natural tail shorter than the other, and an espresso-brown pencil, who had never done her brows and felt she always overdid them. The replies are about the pencil and the method — pluck only stray hairs outside your natural brow shape, comb them with a spoolie, apply in quick short strokes from the beginning of the brow to the tail, and eyebrows are sisters not twins — plus a referral to a professional brow service. Nobody mentions the heart face. August 2024 is the mirror image: the shape is called right as it is and the suggestion is to go a shade or two lighter toward a light-to-medium brown, which a second reply disagrees with outright, saying the natural way the hairs sit suits her.

That is the finding, and it is not a criticism of anyone answering. On this outline the two things that most change how a brow reads are the gap between the brow and the hairline, and the contrast between the brow and the skin around it. The first is section five, and it is the one that costs months to undo. The second costs a wash. It is why two of the thirteen exchanges resolve into a colour recommendation without anyone deciding to, and why the ordering at the end of section seven is the ordering the record keeps arriving at by accident: the reversible edit is the one that gets suggested, because it is the one that can be suggested safely.

The referral, which is the most repeated brow instruction in the pool and the emptiest. Three separate askers with this face were sent to eyebrow mapping — March 2019, to search online for face and makeup maps whenever she had a question about blush, brows or concealer; April 2022, to watch a few videos on brow mapping because it is very helpful; August 2023, to look up eyebrow mapping because it is a huge help. Not one of the three says where the points go. That omission matters more here than it would elsewhere, because the peak point is precisely what section three finds four sources converging on and precisely what the textbook and the measured preference disagree about. A referral to a method whose central landmark is contested is a referral that cannot be followed.

The last one belongs here because it is about the outline rather than the brow, and it closes a loop the page opened. In September 2020 someone described losing five or six pounds and having her face go from round to heart or upside-down triangle, with a more prominent jawline, more prominent cheekbones, visible dimples, and the fat gone from her chin, jaw, neck and cheeks. In April 2023 the person at the centre of the classification fight said she found her own judgement unreliable and could not gauge her face on her own. Two people, three years apart, describing the same thing from opposite ends: the outline moves, and the person looking at it is the worst-positioned observer of the movement.

The same clinical table gives a long face the opposite brow

One document, two rows, and the rows share a measurement. The 2026 review's table defines this outline as a forehead and cheekbones of the same width, a narrower jaw, a more angled chin, long straight jaw lines — and longer length than width. It defines the long face as longer length than width, with straight cheekbones and a squared jawline. The length-to-width clause is verbatim in both. It is the only measure the two rows have in common.

The prescriptions are opposites. For this face: a round brow shape, with the peak between the lateral limbus and the lateral canthus, which softens and balances. For the long face: flat and relatively lower-positioned brows, or soft-angled ones, because they add width to the face's appearance — with an explicit warning to avoid arched and peaked brows, because those create vertical lines that make the face appear even longer. Round against flat. Peak against no peak. Same length-to-width ratio underneath both.

So the brow instruction does not follow from the length measure at all, and the table's own wording shows what it follows from. The difference between the two rows is the jaw: squared on the long face, narrower and more angled here. On the table's logic a bend builds a vertical, a vertical is what a long face already has too much of, and this face has too little of the thing that would balance a wide forehead — which is a jaw, not a brow. The prescription is compensating for the jaw by way of the brow, and the length clause is decoration.

This site's numbers say roughly the same thing, from a different direction, and with the caveat attached. On the detector's constants the length ratio — face length over cheekbone width — is 1.182 for this outline and 1.261 for a long one. Those are 1.65 standard deviations apart, which separates them but not emphatically, and this outline's prototype was interpolated from a single calibrated example rather than measured across a group. The jaw ratio separates them better: 0.759 against 0.803, or 1.69 standard deviations. The forehead ratio separates them worst, at 1.23 standard deviations, 0.978 against 0.951. The clinical table picked the feature that separates them least and the site's constants pick the one that separates them most, and they still reach the same conclusion: the jaw is what makes these two different faces.

Which produces a real problem for anyone sitting between them, and the record contains exactly one such person. The August 2026 asker described a face that is very long, a chin that seems less rounded with age, and cheekbones that seem a little wider, and asked whether she was oval or diamond. She was told heart leaning slightly oval, on the forehead. Take her at her word and both prescriptions are addressed to her simultaneously: round with a peak between the limbus and the canthus, because her forehead is the widest part; flat and low, because her face is very long. They cancel. Neither document anticipates this, and no source in the record mentions the case.

The tiebreak is not a shape name, and this page will not invent one. What it can say is which measurement to trust when the two disagree, and the answer is the one with the larger separation: the jaw ratio, at 1.69 standard deviations, over the length ratio at 1.65. That is a small margin and it is offered as a margin rather than a rule. This site's result page reports both shapes with a margin against each instead of choosing, which is the honest version of the same answer — and the brows are the one feature where you cannot have both, because a bend is either there or it is not.

Brows for a heart-shaped face, in the order the advice was given

Sixteen rows: the thirteen exchanges that name this face and reach the eyebrows, two more that name it and never reach the brow, and one clinical review that prescribes for it by name. Twelve years and eleven months between the first and the last, taken in the order the advice was given rather than the order it was found. The two extra exchanges are here because one is the largest single body of talk about this outline in the record and the other is its chronological last word, and both are marked as carrying no brow instruction. The fourth column is the one to read down: whether the shape of the face was the reason for the brow instruction, which is the test this page opened with. It comes out yes five times in sixteen rows, and only four of those five are exchanges.

Five phrasings a person with this problem would actually type were searched, in fourteen searches: heart shaped face and eyebrows; an eyebrow shape for a heart face; the best brows for a heart shaped face; heart face eyebrows straight or arched; and what eyebrows suit a heart shaped face. To qualify for a row, an exchange had to carry a retrievable date, to contain this face shape, and to contain a sentence about the eyebrows. Thirteen qualified. Three more rows are printed and none of them qualifies: two exchanges name the outline and never reach the eyebrows, and are given rows because one is the record's largest body of talk about this face and the other is its last word in date order; and one entry is not an exchange but a clinical review, given a row because it is the only entry anywhere in the record that is both keyed to this shape by name and precise enough to check. Sixteen rows in total, and the fourth column says which is which. A further set of dated exchanges in the pool name this outline and contain no brow instruction; apart from the two exceptions above they are counted in section eight and given no rows, because a table of what people were told about their brows should not be padded with what they were told about their hair. Three of the largest bodies of brow advice turned up by the same searches contain no heart-face material at all and are excluded outright, so what follows is not the loudest record available, only the on-topic one. One further limitation, stated rather than buried: five of the fourteen searches reached sources outside the pool and returned nothing usable, so the pool is narrower than the five phrasings on their own would suggest, and rows those sources might have added are not here. Sources are described here by who was asked and what was said, and are never linked.

The best brow shape for a heart-shaped face, as far as the record goes

Three positions survive checking, and they are not shape names. The peak goes at about two-thirds of the way along the brow, which four sources reach independently and which on a 60 mm brow is 10.00 mm — 0.39 in — further out than halfway. The top edge of the brow stays where it is, because the gap above it is forehead on this outline and the trade runs between 3.8 and 12.5 to 1 against. And an error costs four and a half to six and a half months, which reorders everything else into set, then colour, then outline, then removal.

Four positions do not survive, and saying which is the useful half. Round versus sharp is unresolved and unresolvable from inside the record, because the two sides are not measuring the same thing — one is a clinical table reproduced from another publication with no measurement attached, the other is one reply in an exchange of hundreds keyed partly to a personality. The five prohibitions on the top edge have no amount in them, and the cost of ignoring that is a factor of three and a quarter depending on your starting height. The mapping referral names a method and no points, at the one landmark the record actually disputes. And the masculinisation question was never answered by anyone, which leaves the only first-person report of a cost attached to this face's standard advice entirely unexamined.

What nobody measured, stated narrowly enough to be checkable. No source in this record reports a measurement of a brow taken against a heart-shaped face. The 2017 study that supplies the 135 degrees and the 7.51 mm never recorded face shape — the words do not appear in it. The 2026 review's table is reproduced from another publication and carries no measurement of its own. The 2016 review states outright that it is impossible to define an ideal eyebrow suitable for every face, and then defines one. Every number on this page comes from a source that was not looking at this outline, and the page's whole argument is that this is still more than the record managed.

One asymmetry to close on, because it explains the shape of everything above. The people asking were specific. They named the shape, the colour mismatch, the tail that grows shorter on one side, the inner half that is dark and straight against an outer half that is curly and sparse, the lamination that lasted a week and then took the hairs with it. The people answering were general. Of thirteen exchanges containing both this face and a brow sentence, four tied the two together, and a fifth statement did the same from inside a clinical review rather than an exchange; two of those four were the writer talking about herself, and the sharpest question in twelve years and eleven months drew zero replies. The record is not thin because the subject is hard. It is thin because the questions were better than the answers, and nobody kept score until now.

• Measure the gap above the brow before touching it, and write the number down. Top of the brow to the horizontal line through the underside of its inner end: 7.51 mm is the measured average across 60 women, the range runs 4 to 13, and one inch is 25.4 mm. On a 50 mm forehead, every millimetre you remove from the top is 2.0 per cent of forehead, and it costs 25.0 per cent of the brow if you started at 4 mm or 7.7 per cent if you started at 13. The instruction is the same for both of you. The price is not.
• Find the peak by fraction, not by name, and check it against a landmark you can see. Two-thirds of the way from the inner start toward the tail. On a 60 mm brow that is 40.00 mm rather than 30.00 — 1.57 in rather than 1.18 in — and it should land between the outer corner of the eye and the outer edge of the iris, closer to neither. A peak over the inner edge of the iris reads as surprised. A low inner end with a high outer peak reads as angry. Both of those are from a source that also says no single ideal exists, so treat them as warnings rather than rules.
• Set the hairs before deciding anything, and decide in that order for the rest: set, then colour, then outline, then removal. A clear gel or a clear wax brushed through with a spoolie changes the reading of the brow without changing the brow, and washes off the same evening. Colour washes off in a comparable time. Removing hair does not, and takes 136 to 204 days to come back.
• Change colour before outline if you are going to change one, and check the pencil against your natural brow rather than your hair. Two of the thirteen exchanges resolved into a colour recommendation without deciding to. In the first, the asker named both her own brow colour and the shade she had been using — taupe brows, an espresso-brown pencil — and the answer was to find a taupe pencil instead, with a second reply suggesting an in-between shade and a third, from someone who also has taupe brows, warning that plain taupe can still read slightly off. In the second, the suggestion was to go a shade or two lighter toward a light-to-medium brown, and a reply disagreed outright, saying the natural way the hairs sat suited her. Neither exchange removed a hair. Neither was wrong, and both are reversible, which is the point.
• If your inner half is darker and denser than your outer half, the front is the end to take hair from, not the top and not the bottom. Plucking a few hairs from between at the inner start reduces the front's density so it matches the tail, and it is the only instruction in the record that addresses that specific mismatch. The inference after it is this page's own, and no source in the record makes it: that mismatch is what this outline's own definition creates, because a narrower jaw and a tapered chin leave the outer half of the face less to balance against.
• Take photographs, because that is the only reversal mechanism the record contains. Small changes, photographed, compared — offered in November 2020 by someone advising a person with this face, and independently reported in August 2024 by someone who had paid a technician, been annoyed at how little was done, and then used the photographs as her guide for years. Both arrived at the same method without citing anything. Neither is a measurement, and on a subject where four and a half months is the smallest unit of feedback, it is the only instrument anyone has.