Measure your pupillary distance to the half millimetre

What pupillary distance actually is

Pupillary distance, almost always shortened to PD, is the distance in millimetres between the centres of your two pupils. It exists on a prescription for one reason: every corrective lens has an optical centre, and that centre needs to sit directly in front of the pupil it is correcting. Get it right and you look through the part of the lens the prescription was calculated for. Get it wrong and you look through a slightly off-axis part of the lens instead, which introduces unwanted prism.

For a mild prescription the effect is small enough that most people never notice. As power increases, the tolerance narrows quickly, and a few millimetres of error starts producing eye strain, headaches, or the peculiar sensation that the floor is not quite where it should be. This is why PD matters far more for a minus-six lens than for a minus-one, and why it matters most of all for progressives, where the corridor you read through is only a few millimetres wide.

Single PD and dual PD

Two conventions exist, and retailers ask for them inconsistently.

Single PD, also called binocular PD, is one number describing the full pupil-to-pupil distance. It is what most online checkout forms want, and it is the number you should paste in if the form has a single box.

Dual PD, also called monocular PD, is two numbers, each measured from the centre of the bridge of the nose to one pupil. It is conventionally written right eye first, as in 32/30. Dual PD exists because faces are not symmetric: it is completely normal for one pupil to sit a millimetre or two further from the midline than the other. Splitting a single PD down the middle would put one optical centre slightly off on both lenses. Progressive lenses are fitted to dual PD for exactly this reason.

This tool reports both. It derives the split by projecting each pupil onto the line joining them, measured from the midpoint between your inner eye corners. That choice matters more than it sounds. The obvious midline reference is the bridge of the nose, and it is the wrong one: the bridge sits a centimetre or more in front of your pupils, so the moment your head turns a few degrees, perspective drags it sideways relative to the pupils and the split becomes wildly asymmetric. The inner eye corners sit at pupil depth, so they move with the pupils and the split stays honest. Even so, the dual figures are the first thing a head turn corrupts, and the tool tells you when that is happening.

Why your optician may not have given you the number

This trips up an enormous number of people ordering glasses online for the first time. You have a valid prescription in your hand, the retailer asks for your PD, and the prescription simply does not contain it.

That is not an oversight. In the UK and in most of Canada, pupillary distance is classified as part of dispensing rather than part of the sight test. The prescription records what your eyes need; the PD is a fitting measurement taken when the frames are made up. An optometrist therefore has no legal obligation to write it down for you, and a practice that also sells frames has little commercial reason to volunteer it. In the United States, practice varies by state and by clinic.

You can always ask, and most practitioners will measure it for you on request, often free of charge. If asking is awkward or you have already left, measuring it yourself is a perfectly reasonable alternative for a standard single-vision order.

How this tool measures it

A camera gives you pixels. Turning pixels into millimetres needs a reference object of known size somewhere in the image. Most free PD tools solve this by asking you to hold a bank card against your forehead. That works, but it has a flaw nobody mentions: the card sits a centimetre or two in front of your pupils, and because a camera is a perspective projection, objects nearer the lens cover more pixels per millimetre. A card-referenced measurement therefore reads slightly large unless the error is corrected.

It has a second, more human flaw. Asking someone to photograph themselves holding a bank card is asking them to photograph their card number, and no assurance about local processing makes that a comfortable request. Any card built to the ID-1 standard is exactly 85.60 mm on its long edge, so a loyalty card, gift card, transit pass, hotel key or long-expired card serves identically — and you can always turn the card so its plain side faces the lens.

The primary method here uses a different reference: your own iris. The horizontal visible diameter of the adult human iris is remarkably consistent, sitting close to 11.7 mm with roughly half a millimetre of population spread, and it barely changes after childhood. Crucially, the iris lies in the same plane as the pupil, so there is no perspective error at all. You need no props, and the reference cannot be held at the wrong angle.

The card option remains available, because two independent methods that agree are worth more than one method asserted confidently. Whenever a card is marked up, the tool computes the iris-referenced figure too and shows you both, along with the gap between them. If they disagree by more than about a millimetre and a half, something in your setup is off, and you want to know that rather than be handed an average.

One consequence of using the iris as the reference deserves stating plainly: take your glasses off. A corrective lens bends light before it reaches the sensor, moving the apparent position of your iris and changing its apparent width. That corrupts the scale and the measurement at the same time, in the same direction, so the error does not announce itself.

Rejecting bad frames instead of averaging them in

A turned head is the single biggest source of error in self-measured PD. Rotate your face fifteen degrees away from the camera and your pupil separation foreshortens by about three percent, which on a 63 mm PD is nearly two millimetres lost, silently.

The tool reads head pose from the face model's transformation matrix on every frame and throws away any frame where yaw exceeds eight degrees, pitch exceeds twelve, or your head is rolled more than twelve degrees off vertical. It also checks that both eyes are open far enough relative to iris size, so a blink cannot contribute, and compares the two iris diameters against each other, since a large disagreement is the signature of a rotation the pose estimate may have missed.

When we refuse to answer

Some photographs cannot be measured, and the useful thing to do is say so rather than produce a confident number. Sunglasses are the clearest case: the face model will place landmarks behind a dark lens quite happily, and the result looks entirely plausible. The giveaway is that the eyelid aperture collapses relative to the iris, which tells us the model is guessing. In that situation the tool reports nothing at all, and does not draw a figure on the image either.

A head turned more than fourteen degrees, a chin tilted past eighteen, or a figure outside the range a human face can produce are treated the same way. A milder tilt is different: the reading is shown, with a warning attached, because refusing a slightly imperfect photo helps nobody.

Making the three figures agree

Opticians dispense to the half millimetre, so that is what we display. Rounding the single PD and the two monocular values independently lets them contradict each other, giving you 30.5 and 32.0 beside a single PD of 62.0. We therefore round the total once, round one side of the split, and derive the other side from the total. Each figure stays within a quarter of a millimetre of what was measured, and the two monocular values always add up to the number printed directly above them.

Reporting a range, not a single number

Across roughly sixty accepted frames the tool takes the median rather than the mean, so one bad frame cannot move the answer, and publishes the interquartile spread as a confidence range. This is the honest way to present a camera measurement. If your frames disagree by two millimetres, a single tidy number would be a lie of presentation, and you deserve to see the scatter so you know whether to retake.

Getting the best possible reading

• Use even, frontal light. A window in front of you is ideal; a bright light behind you is the worst case.
• Sit an arm's length away, with your whole face filling most of the frame.
• Look directly at the camera lens, not at your own image on the screen. These are a few centimetres apart and the difference is measurable.
• Keep your head level and square. Chin down or turned even slightly will shorten the reading.
• Remove glasses. Frames and lens distortion both interfere with iris detection.
• Push hair off your brow so the nasion is visible for the dual-PD split.
• Measure two or three times over a minute. If the answers agree to within half a millimetre, you have your number.

Near PD, for reading glasses

When you focus on something close, your eyes converge and your pupils move slightly closer together. Reading glasses are therefore fitted to a near PD, which is smaller than your distance PD. The conventional adult adjustment for a working distance of about forty centimetres is to subtract three millimetres, and that is what the near figure above shows. If a retailer asks specifically for near PD, use that number; if it asks simply for PD, use the distance value.

Typical values

Most adults measure between 54 and 74 mm. Averages land near 63 mm for men and 60 mm for women, with wide overlap. Children measure considerably smaller and the value grows until the late teens, which is why children's frames should always be measured rather than estimated from an adult average. If your result lands well outside the usual adult band, the most likely explanation is a measurement problem rather than an unusual face, so retake it before ordering.

What goes wrong when PD is wrong

An incorrect PD does not usually produce dramatically blurred vision. It produces something more insidious: mild, persistent discomfort that people tend to blame on the prescription. Eye strain after an hour of reading, a dull ache behind the brow, words that seem to swim slightly at the edge of the page, or a sense that the lenses are fighting you. Lens misalignment is a well-documented contributor to online eyewear returns, and it is entirely avoidable.

If you already own glasses that feel wrong and you suspect the fit, it is worth measuring your PD and comparing it against what was ordered.

When to see a professional instead

Measure it yourself with confidence for single-vision distance or reading glasses in an ordinary prescription range. Have a dispensing optician do it if you are ordering progressive or bifocal lenses, if your prescription is strong in either direction, if there is significant prism in your prescription, if you are buying for a young child, or if you have any condition affecting eye alignment. In those cases the fitting involves more than PD alone, and a few millimetres genuinely matter.

This tool provides a measurement, not eye care. It does not diagnose anything and is not a substitute for an eye examination or professional dispensing advice. If you are experiencing changes in your vision, see an optometrist or ophthalmologist.

Questions people ask

Is measuring my own pupillary distance accurate enough to order glasses?

For single-vision lenses in ordinary prescriptions, a careful self-measurement is generally close enough, and it is what online retailers expect when your prescription omits the value. For progressive, bifocal or high-power lenses, where a millimetre of misalignment is noticeable, have it measured by a dispensing optician.

Why is my pupillary distance not written on my prescription?

In the UK and most of Canada, pupillary distance is classified as part of dispensing rather than part of the sight test, so an optometrist has no obligation to include it. In the United States practice varies by state and by practice. You can always ask for it, and most will provide it.

What is the difference between single PD and dual PD?

Single or binocular PD is one number: the distance between the centres of your two pupils. Dual or monocular PD is two numbers, each measured from the centre of the bridge of your nose to one pupil. Dual PD is more precise because most faces are slightly asymmetric, and it is what progressive lenses are fitted to.

Does this tool upload my photo?

No. The face model runs inside your browser using WebAssembly. No image, video frame or measurement is transmitted anywhere. You can confirm this by opening your browser's network tab while you measure.

What is a normal pupillary distance?

Most adults fall between 54 and 74 mm, with adult averages around 63 mm for men and 60 mm for women. Children measure smaller and the value increases until roughly the late teens, which is why a child's frames should be measured rather than estimated.

Do I need a credit card to measure my PD?

No. The iris method needs no props at all: it uses the width of your own iris as the reference, which sits in the same plane as your pupils and so introduces no perspective error. If you do want the card cross-check, any card of the standard ID-1 size works — a loyalty card, gift card, transit pass, hotel key or expired card are all exactly 85.6 mm wide. You never need to show a card number, and you can turn the card so its plain side faces the camera.

Should I take my glasses off before measuring?

Yes. A prescription lens refracts light before it reaches the camera, which shifts where your iris appears and changes how large it looks. Since the iris is the millimetre reference, measuring through lenses introduces an error in both the scale and the position. Remove them for every method.

Why does the tool sometimes refuse to give a confidence range?

Because a range needs enough frames to mean something. A single photograph has no frame-to-frame scatter to measure, so we report that no range could be established rather than printing a tidy plus-or-minus we have not earned. Run the camera measurement to get one.