How to Measure Glasses Frame Size: Four Rulers, One Subtraction, and the Width It Pays

Four rulers, and the four segments they touch

Every number in a published size chart is taken with one of four rulers, and the charts rarely say which. A sizing page published 27 April 2023 and revised 29 June 2023 sorts frames into five bands — 100 to 126.9 mm, 127 to 130.9, 131 to 139.9, 140 to 143.9 and 144 to 180 mm, or 3.94 to 7.09 in across the ladder — and calls what it sorts the overall frame width, a figure that page says its retailer prints on every product page. A frame fitting guide published 16 June 2025 and updated 28 July 2026 sorts 125 to 140 mm (4.92 to 5.51 in) into four bands and is explicit about where the ruler goes: the frame laid flat, measured horizontally from hinge to hinge. An undated consumer guide sorts frames by the width of a single lens instead — 50 mm (1.97 in) or less, 51 to 54 mm, and anything wider than 55 mm (2.17 in). And a fourth undated page skips the frame altogether and asks you to stand in front of a mirror with a ruler against your own face.

The four segments differ by more than the spread of any chart that uses them. Take the most ordinary size in the most common example, a front stamped 52□18. Its lens centres sit 70 mm (2.76 in) apart, because that is the first number plus the second, and a boxing-system reference published 10 June 2026 and revised 2 August 2026 gives exactly that sum as its worked case. Its hinge-to-hinge span is smaller than its end-piece-to-end-piece span by whatever each end piece runs past the hinge — a quantity no frame prints. And its eye size, 52 mm, is a third of the front a shopper is actually comparing against a chart. Four rulers, four numbers, and the frame itself carries only the smallest.

None of which makes the charts wrong at what they are for. Three of the four answer a question a shopper genuinely has: will this look proportionate, and will it sit where the pair I own sits. The fourth question — is this width centred enough that somebody can put my prescription into it — is not answered by any of them, and it is not a taste question. It has one input that none of the four ladders takes.

The three numbers on the inside of a temple arm are the lens width, the bridge width and the temple length, in millimetres, and this site already has a page that reads them — what that stamp settles on its own, and what it leaves open. What that page does not do is start from your eyes and work outward. That direction is the subject here.

The subtraction, run backwards

Two halves of one calculation are published and the third is not. The first half is a sum: add the eye size and the bridge and you have the distance between the two geometric lens centres, which the 10 June 2026 boxing-system reference above writes as Frame PD = A + DBL and works as 52 + 18 = 70 mm (2.76 in). The second half is a halving: the distance each lens has to travel off its geometric centre is half the difference between that sum and your own pupillary distance, which a lens-laboratory technical guide published 9 March 2026 and revised 17 September 2026 prints as Decentration per eye = (Frame PD − Patient PD) / 2, and whose consequence it names in one direction only — excessive decentration requires larger blanks, increasing thickness and weight.

Nobody in the consumer market runs it in the other direction. In seven readable pages at the top of this question, the word decentration appears zero times, and no page solves the first number from the second. The same identity solved for the lens width, though, is the only way the two published ceilings become a purchase decision rather than a laboratory note:

lens width = pupillary distance + (2 × the decentration you will accept) − bridge width

Two ceilings are in print, both somebody else's published figures, and they do not agree. The 9 March 2026 laboratory guide states it as a general guideline to keep decentration under 5 mm per eye where possible, by choosing frames with a pupillary distance close to yours. A styling guide published 3 April 2026 puts the target tighter and in a band — for most wearers, the goal is to keep decentration under 3–4 mm per eye when possible, though exact tolerances depend on your prescription strength and lens design. The table below is computed at three of them: 5 mm (0.20 in), and the two ends of that band, 4 mm (0.16 in) and 3 mm (0.12 in). The 3–4 mm band is published by a page about powers and lens designs, so the two right-hand columns are where higher prescriptions live; nobody should read them as a fit rule for a pair of readers bought off a width chart.

One bridge value is held down the whole table: 18 mm (0.71 in), a mid-range inside the published adult bridge band of 14 to 24 mm (0.55 to 0.94 in), so that the only thing moving between rows is your pupillary distance. Section four runs the same row with the bridge as the variable instead. The pupillary distances are the published adult band's own edges and interior — 54 to 74 mm (2.13 to 2.91 in), per the 3 April 2026 styling guide — and the 63 mm row sits in the middle because that is the distance figure a prescription-lens retailer uses as the input to its own worked example; that page does not label it an average, and neither does this one.

Read column five before using the other three. A 54□18 front puts its lens centres 72 mm (2.83 in) apart; a pupillary distance of 63 mm is 9 mm narrower than that, so each lens has to travel 4.5 mm toward the nose — that same row, with that same direction, is already printed on this site in the guide to the numbers stamped on the arm, which works forward from the stamp rather than backward from the eyes, and this page does not recompute it. Above a 72 mm pupillary distance the column changes meaning: the front is now the narrower thing, and the lenses travel outward instead. At the bottom edge of the published adult band a 54□18 front asks for 9 mm per eye, which is over the looser ceiling by four millimetres on a frame nobody would describe as unusual. And note where columns two and three leave the published lens band: the 68 mm row already reaches 60 mm (2.36 in), the top of the printed 40 to 60 mm range, and everything above it escapes the range unless the bridge is taken to the wide end of its own band.

How to measure glasses size with a ruler, and where that leaves the lens

The ruler route is real, and it is the only way to order a replacement pair without trying anything on. One eye care practice's fitting guide, published 23 August 2022, tells you to hold the ruler horizontally across the face just below the eyes and note the distance between the left and right temples. A contact-lens and eyewear retailer's guide, published 25 September 2023 and revised 11 October 2023, tells you to start at the outer edge of one temple and run the ruler level across the front of the face to the outer edge of the other. Both are measuring a face. Then the second of them says the result is your eyeglass width or frame width, and adds that many frames carry a figure printed on the arm as "frame width" or "eye size" — which is where the route breaks, because the number on the arm is the eye size, one lens, and a temple-to-temple face reading is a whole front. A 135 mm face reading and a 54 on an arm are not two attempts at the same measurement.

The route works anyway, with one correction: treat the face reading as a constraint on the front — the front is genuinely what a face measurement constrains — and get the lens width from the subtraction instead. A frame that presses leaves marks at the temples; a frame much wider than the face rides down the nose, which is the failure the 23 August 2022 fitting guide warns about in its own words. Neither of those is a statement about where the lens centres land. So take a 63 mm pupillary distance against the 5 mm ceiling: the width that pays is 55 mm (2.17 in) with an 18 mm bridge: a 73 mm sum, and a front of roughly 148 mm (5.83 in) with an ordinary end piece. On a face measuring about 135 mm that front is generous, and the dimension that has to give is the bridge, not the width — take the bridge to the top of its published band at 24 mm (0.94 in) and the same ceiling on the same face pays 49 mm (1.93 in) of lens and a front near 142 mm (5.59 in).

That is also how a face measurement can look entirely reasonable and still land on a lens a laboratory will not cut. A ruler cannot see the sum. A front chosen because its width matched a face can carry two lens centres 75 mm (2.95 in) apart, which on a 63 mm pupillary distance is 6 mm per eye and outside both published ceilings, and the only two numbers that describe it — the lens width and the bridge — are sitting on the arm in a form that has to be added rather than compared to anything.

Hold the bridge steady, and the frame pupillary distance stops being 72

Solve the identity for the lens width while the bridge stays fixed and something counter-intuitive happens: the frame pupillary distance, the quantity the whole published apparatus is organised around, stops moving. At a 63 mm pupillary distance and 5 mm per eye the two lens centres are 73 mm apart whatever the bridge is — 59 + 14 = 73, 57 + 16 = 73, 55 + 18 = 73, 53 + 20 = 73, 51 + 22 = 73, 49 + 24 = 73. Bridge and lens trade one millimetre for one, and the sum never notices. What does move is the front, from 152 mm at the narrowest bridge to 142 mm at the widest, and that is not a cosmetic difference: it is the difference between a frame that ends at your cheekbones and one that does not.

Change which quantity is pinned and the sensitivity inverts. Pin the sum at 72 mm — the 54□18 example every retail guide prints — and the bridge now sets the lens width by simple subtraction: 54 mm at a 18 mm bridge, 58 at 14 mm, 50 at 22 mm. And a 72 mm sum is only the right answer for one face: at 5 mm per eye it stops fitting a pupillary distance below 62 mm (2.44 in); at the 3 mm end of the published band it needs 66 mm (2.60 in) or more before the travel is inside it. The published adult band's own lower edge, 54 mm, is 9 mm per eye away from that example front, which is nearly twice the looser ceiling.

Below, the same 63 mm row runs with the bridge as the variable, and it is where the four rulers of section one and this arithmetic meet. The sum column is flat; the front column is not. Two of the retail ladders quoted at the top of this page sort on the front — one end piece to end piece, one hinge to hinge — and the third on the single lens, which is column two.

The 20 mm and 12 mm allowances are this page's own numbers, the pair total for both end pieces, and they exist in the table for one reason: to show sensitivity. Change the allowance and only the three columns containing it move — column two and column three do not know it exists, which is the whole difference between a quantity the frame publishes and a quantity a shopper has to estimate. Four of the six rows fall inside the 125 to 140 mm hinge-to-hinge ladder, and the two rows with the widest bridges fall just below it; none of the six is anywhere near the 73 mm sum in column three, which is the figure the lens width in column two was actually solved against. Column four at 152 mm (5.98 in) also sits above the 120 to 150 mm front range published on 3 April 2026 — two millimetres over, on a frame that is inside the published lens band and the published bridge band at the same time.

The tolerances, in both directions

A sizing page published 27 April 2023 and revised 29 June 2023 gives the most usable tolerance rule in print, and it is written for shopping rather than for optics: the bridge width can be within 2 mm (0.08 in) of your current frames and still provide a proper fit, and all other measurements can be within 2 to 3 mm of the corresponding dimension on the frames you already have. Nothing on that page mentions a pupillary distance. Read against the identity, the rule spends something it does not name.

The rule quoted above is the closest thing in print to an answer for someone who arrived asking how to measure your glasses frame size, because it is the only published instruction in this set that says how far off you are allowed to be. Everything else in the published material hands you a number and calls it exact. Let the bridge move by the 2 mm the rule grants while the lens stays where it is, and the frame pupillary distance moves 2 mm with it, which is 1 mm per eye of decentration you did not ask for and did not order — a fifth of the laboratory guide's ceiling, and a third of the tighter end of the styling band. Hold the decentration still instead, which is what the tables above do, and that same 2 mm of bridge costs 2 mm of lens width, one for one, in the other direction. Run the rule to the ends of the published bridge band on a single pupillary distance and the tolerance buys 10 mm of lens width, 49 mm against 59 mm — half the span of the published 40 to 60 mm lens band, moved by the dimension a frame prints second.

Those two published ceilings disagree with each other by about the same amount, and the disagreement is arithmetic-sized rather than editorial. At a 63 mm pupillary distance and an 18 mm bridge, the 5 mm reading and the 3 mm reading differ by 4 mm of lens: 55 mm (2.17 in) against 51 mm (2.01 in). Four millimetres is more than the tolerance a shopper is told to allow on any dimension except the bridge. The 3–4 mm band carries its own qualifier on the page that published it — the tolerance depends on prescription strength and lens design — so the honest reading is that the 5 mm column is the one a width chart can support, and the tighter columns are somebody else's decision about your powers.

One more substitution is worth having ready, because it moves the answer without changing anything about the frame. The 9 March 2026 laboratory guide also gives the near-work version of the identity, decentration per eye set against the near pupillary distance, and states that the near figure is typically smaller than the distance measurement by 3 to 4 mm in total, about 1.5 to 2 mm per eye. A prescription-lens retailer's own worked example uses 63 mm at distance and 60 mm at near. Run 60 mm through the same formula with the same bridge and the same 5 mm ceiling and the answer is 52 mm (2.05 in) — the width, the bridge and the sum of a 52□18 front, which is the size a laboratory page and a boxing-system reference both reach for as their illustrative case. It is not a coincidence that 52□18 is everybody's example: at a near pupillary distance it is exactly what the looser published ceiling pays.

The page that adds a front, and the page that names a lens by two names

Two of the consumer pages this one disagrees with are disagreeing with themselves, and both failures are checkable from the page in front of you. An undated buyer's guide tells a shopper to stand before a mirror, hold a ruler in line with the temple, measure the distance between the temples in inches and multiply by 25.4 to get millimetres. Then, in the same list of steps, it says the frame width you want is the sum of both lens widths plus the bridge width, and that the frame size may vary from your measurement by up to 3 mm (0.12 in). Further down the same page it sorts frames for narrow faces below 129 mm (5.08 in), medium faces at 130 to 139 mm, and wider faces above 139 mm.

Now hold those three instructions against one another at the width this page has been computing. A 63 mm pupillary distance with an 18 mm bridge at the 5 mm ceiling pays a 55 mm (2.17 in) lens. The ruler-on-the-face route is a face measurement, and the add-the-parts route applied to those same parts gives 128 mm (5.04 in) — two lenses and a bridge, with no end pieces, because no end piece was mentioned. The object those same parts actually make is a front near 148 mm (5.83 in) once each end piece runs 10 mm past the hinge. On that page's own ladder the first number is what it prescribes for a narrow face and the second is what it prescribes for a wide one, twenty millimetres apart, from the same three inputs and the same frame. The ±3 mm tolerance does not reach the gap; the gap is exactly the two end pieces the addition drops, and it is a gap in the definition of a front, not in the arithmetic.

A second self-conflict is smaller and more common. A fitting guide published 25 September 2023 and revised 11 October 2023 measures the face from the outer edge of one temple to the outer edge of the other and then names that number, in the same breath, your eyeglass width or frame width — and notes that the figure printed inside the arm is often labelled either "frame width" or "eye size". Those two labels are roughly ninety millimetres apart on any adult frame: the arm carries a single lens, 50 to 56 mm, and the ruler has just taken a front. That is not a mistake a reader can correct from the page, and it is the reason a face measurement keeps turning up in charts that are sorting frames. What the arm actually publishes is the pair of numbers that sum to the frame pupillary distance; the front has to be estimated, and the face has to be measured, and neither of those is the number on the arm.

Twelve published width positions, and what each one sorts on

Nothing below is a criticism of the pages that published it, because each of these positions was written to answer a different question than the one this page was built around. The table is sorted by the date each source carries, oldest first, with one thing quoted per source: the sentence about widths, not a summary of the page. Where a source carries a publication date and a later revision date, both are given; where the ladder itself sits on a page that carries no machine-readable date, the row says so rather than borrowing a date from a neighbouring page. The span from the oldest dated position to the newest is 2,796 days — seven years, seven months and 27 days, and this page does not round it up.

The column worth reading is the last one. A reader looking for a glasses frame size chart will find twelve of them in print and they sort on five different widths: one lens, the whole front, the hinge span, the face, and the blank a laboratory happens to stock. A sixth quantity appears in the list exactly once — a pupillary distance — and that row is not a width chart at all. That is the whole of the gap this page is filling, stated without claiming that anybody printed a wrong number.

Five of the twelve sort a frame by a width and never take a pupillary distance as an input; one takes the pupillary distance and never sorts anything by width; two print both quantities and never join them; and one page holds two recipes for the same measurement that disagree by exactly the end pieces one of them omits. Between the oldest dated row and the newest, 2,796 days, nothing in this list was corrected — it was accumulated, which is a different thing and the reason the arithmetic still has to be done by the shopper.

What to do with a pupillary distance you already have

Three numbers and one subtraction, in this order. Take the distance pupillary distance off a prescription rather than off a mirror — an optician's measurement, and if the sheet carries a near figure as well, use the distance one for distance glasses and expect the near one to be 3 to 4 mm smaller in total. Then take the bridge width off the frame you already wear comfortably, which is the second number on the arm, and treat 2 mm either side of it as room the published tolerance rule grants you rather than as adjustment that is free. Then add: lens width plus bridge is the frame's pupillary distance; subtract yours; halve what is left. If that figure is over 5 mm per eye the width has to come down, and if the powers are high the target worth working to is the 3–4 mm band, which takes another 2 to 4 mm off the lens at that bridge. The number that comes out will usually not be stamped on a frame you can buy, because frames are sold as combinations and nobody sells a solved width. That is not a defect in the subtraction: it turns the figure into the middle of a band to shop inside, and the tolerance rule in section five is what decides how far off it you are willing to land. The site's own face-shape calculator measures the face, which is a different job and the one a ruler is actually good for: it tells you what the front has to be, never what the sum is.

One row of this page's own arithmetic is worth carrying, because it sits exactly between the two published ceilings and both kinds of chart. At a 63 mm pupillary distance and an 18 mm bridge, the 5 mm reading pays 55 mm (2.17 in) of lens and the 3 mm reading pays 51 mm (2.01 in). The two frames differ by 4 mm of lens, 4 mm of sum — 73 mm against 69 mm (2.87 in against 2.72 in) — and by roughly 8 mm of front, 148 mm down to 140 mm. Both widths are inside the published 40 to 60 mm band; both bridges are inside the published 14 to 24 mm band; every retail chart in the table above will happily sell you either one, and the only thing that separates them is a distance measured between two eyes.