Original analysis · Anthropometry
Is standard desk height too tall? What 6,068 measured bodies say
Almost every desk-height guide — including this site — repeats some version of “your desk should be about 40% of your height.” It is a decent rule and it is also biased. We took a public dataset containing individual body measurements for 6068 adults, computed the quantity the rule is trying to estimate, and compared the two. This page is that comparison: the numbers, the chart, the method, and the parts where we do not think the data supports a strong claim.
Definitions
What we actually measured
The number a desk height rule is trying to hit is your sitting elbow height: floor to the underside of your forearm when you sit upright, shoulders relaxed, elbows bent to about 90°. Put the desktop there and your wrists stay straight without shrugging.
The source is the 2012 Anthropometric Survey of U.S. Army Personnel (ANSUR II), technical report NATICK/TR-15/007: 4,082 male and 1,986 female subjects, 93 direct measurements each, released for unlimited public use. Unlike most summaries, this release includes individual-level records rather than just population averages, so we could compute the distribution instead of guessing at it.
The unit trap, stated up front. In this dataset, sitting elbow height and sitting eye height are recorded above the seat surface. The desk height you want is measured from the floor. Getting this wrong by one line of code produces nonsense — we did exactly that on the first run and got a 445 mm “error.” Every figure on this page is floor-to-target, and the script at the bottom shows the conversion.
Test 1 · the formula
The 0.40 rule overshoots by about 3 cm
Take all 6068 people, compute 0.40 × stature for each, subtract their measured sitting elbow height. Positive means the rule told them to sit at a desk higher than their body wants:
| Error of 0.40 × height | Share of the sample |
|---|---|
| within ±10 mm | 17.1% |
| within ±20 mm | 34.2% |
| within ±30 mm | 51.6% |
| within ±50 mm | 82% |
Median error +29 mm, standard deviation 24 mm, 5th to 95th percentile -10 to +68 mm. Only about half the sample lands within 3 cm of what the rule tells them.
Figure 2 · error of the 0.40 rule
The important part is not the size of the error, it is the shape. The peak sits clearly to the right of zero, and the bias holds steady across every height band in the table below — roughly +25 to +30 mm everywhere. Random error averages out; a bias that is the same at 155 cm and at 190 cm is a coefficient that is simply a little too large. The measured median ratio is 0.3832, not 0.40 — about 29 mm lower at 175 cm.
The other two coefficients survive the same test much better. Chair height (popliteal ÷ stature) measures 0.2428 against the 0.25 we use, and seated eye height measures 0.7016 against 0.70 — a 3 mm difference, which is less than one notch on a monitor arm.
Test 2 · the market
What people need vs what you can buy
Measured sitting elbow height across 6068 adults, barefoot: 587 mm at the 5th percentile, 657 mm at the median, 726 mm at the 95th. Add 25 mm for typical footwear and the median becomes 682 mm.
Figure 1 · desk height the body wants
Set that against what is actually sold. Fixed-height office desks cluster around 730 mm (roughly 29 inches); 710, 740 and 750 mm are also common. Counting anyone whose required height is more than 20 mm off as “not well served”:
| Desk height | Too high for | Fits within ±20 mm | Too low for |
|---|---|---|---|
| 710 mm | 56% | 30% | 14% |
| 730 mm | 73% | 22% | 5% |
| 740 mm | 80% | 17% | 3% |
| 750 mm | 86% | 12% | 2% |
We show both versions because the barefoot figure is the one that gets challenged. With shoes — the condition a desk is actually used in — a 730 mm desk is too high for 73% of this sample and fits 22% within 2 cm. The barefoot numbers are harsher (89% and 10%) and we think the shod ones are the fairer read.
Note what we are not saying. We are not saying standard desk height is “wrong.” Desks are built for a population, not a person, and 730 mm is a reasonable compromise for a sample whose median requirement is 682 mm only if you accept that most people will be working above their elbow height. What the data shows is that the compromise lands above the majority, which is worth knowing before you buy a desk that cannot move.
Breakdown
By height band
Real median desk height for each 5 cm slice of stature, next to what 0.40 × height would tell you:
| Height | n | 5th pct | Median | 95th pct | 0.40 × height | Error |
|---|---|---|---|---|---|---|
| 150-155 cm | 174 | 550 mm | 588 mm | 617 mm | 613 mm | +25 mm |
| 155-160 cm | 478 | 566 mm | 602 mm | 642 mm | 632 mm | +30 mm |
| 160-165 cm | 792 | 583 mm | 622 mm | 662 mm | 650 mm | +28 mm |
| 165-170 cm | 1046 | 600 mm | 640 mm | 681 mm | 670 mm | +30 mm |
| 170-175 cm | 1313 | 622 mm | 660 mm | 704 mm | 690 mm | +30 mm |
| 175-180 cm | 1149 | 640 mm | 681 mm | 720 mm | 710 mm | +29 mm |
| 180-185 cm | 715 | 657 mm | 700 mm | 739 mm | 728 mm | +28 mm |
| 185-190 cm | 266 | 677 mm | 718 mm | 757 mm | 746 mm | +28 mm |
| 190-195 cm | 74 | 694 mm | 740 mm | 780 mm | 768 mm | +27 mm |
The error column barely moves. That consistency is the whole argument: the rule is not failing at the extremes, it is uniformly about 3 cm optimistic.
The ceiling
Why no single formula can fix this
Height explains 69% of the variation in required desk height (R² = 0.693). The remaining 31% is body proportion: two people of identical stature can differ by 6 cm or more in sitting elbow height because one has a long torso and short legs and the other the reverse.
That is a ceiling, not a tuning problem. No coefficient — 0.40, 0.3832 or anything else — gets past it, because the missing information is not in the input. Which is why the honest recommendation is to measure yourself once: sit against a wall, elbow at 90°, measure floor to elbow. It takes thirty seconds and it beats every formula on this page.
Sex is the one input we checked and left out. Male median desk coefficient 0.3843, female 0.3809 — under half a percentage point, about 6 mm at 175 cm. A gender field would buy less than one adjustment notch, so the calculator still only asks for height.
Action
What to do with this
- Measure, then adjust. Sitting elbow height is the number; the formula is only a starting point when you have no tape measure handy.
- If the desk is too high and it adjusts, drop it and re-check the chair underneath — chair height follows desk height, not the other way round.
- If it does not adjust, the fixes are a lower keyboard surface, a chair that goes high enough plus a footrest, or a desk that moves. Which one depends on how far off you are.
- Check the screen separately. Seated eye height is a different measurement with a different coefficient, and the top of the screen goes there — not the middle.
- Footrests only fix feet. If your feet dangle, a footrest helps; if your elbows are above the desk, it does nothing. Here is how to tell which one you have.
Reproducibility
Method, and how to reproduce it
Every number above comes from one script and one public dataset. If you want to check it — or prove us wrong — the whole path is here:
- Download ANSUR II individual records (public release, NATICK/TR-15/007): 4,082 male + 1,986 female = 6068 subjects.
- Convert to floor-based targets:
desk = poplitealheight + elbowrestheight,chair = poplitealheight,eye = poplitealheight + eyeheightsitting. - Compare against 0.40 × stature, tabulate the distribution, and count coverage for fixed desk heights at ±20 mm.
Free to reuse. The derived dataset, the script and every chart on this page are released under CC BY 4.0. If you cite it, attribution to yangmk at yangmk.com/research/standard-desk-height/ is enough. We would rather be corrected than agreed with — if you rerun this and get different numbers, tell us.
Scope
Where this does not apply
- The sample is not office workers. These are U.S. Army personnel measured in 2012, two-thirds male, all of them meeting a physical fitness standard. What travels well is the relationship between measurements — how limb and torso proportions vary at a given stature — not the absolute centimetres. Treat the shape of the findings as more reliable than the exact values.
- Measurements are barefoot. We apply a +25 mm shoe allowance for the desk comparisons. It is an assumption, not a measurement; heel heights vary.
- Sitting elbow height is a proxy for desk height. It is the standard one, but the keyboard surface is often set 1–3 cm lower than the desktop, and thickness of the desk itself is not part of any of these numbers.
- This says nothing about desks for standing. Standing elbow height is a separate measurement with a separate coefficient.
- None of this is medical advice. Geometry can reduce mechanical strain. Persistent pain, numbness or weakness belongs with a clinician.
Questions
FAQ
Is a 730 mm desk too high?
For most people in this sample, yes — 73% of the 6068 adults measured have a sitting elbow height more than 20 mm below 730 mm once shoes are counted. That does not make 730 mm a defective product; it makes it a compromise that lands above the median person. If your desk does not adjust, measure your own elbow height before assuming you are the exception.
Should I use 0.3832 × height instead of 0.40?
You can, and on this dataset it will be closer for the median person — by about 3 cm. It will still miss roughly half the population by more than 3 cm, because the dominant error is individual body proportion, not the coefficient. Measuring your own elbow height once beats any coefficient.
Why not just lower the chair?
Because the chair has to keep your feet on the floor. Dropping the seat to get your elbows under a high desk leaves your thighs unsupported and your feet dangling, and a footrest fixes the feet but not the elbow-to-desk gap. When the desk is the wrong height, the desk is the thing to change.
Does sex matter for desk height?
Barely. Male and female median desk coefficients differ by about 0.003 — roughly 6 mm at 175 cm, which is smaller than one adjustment step on most chairs. That is why the calculator asks only for height.
Can I use these numbers or the chart in my own article?
Yes. Everything on this page is CC BY 4.0 — quote the figures, embed the chart, or rerun the analysis with the CSV and script linked above. Attribution to yangmk with a link is all we ask.
Get your own number in 30 seconds
Enter your height — or your measured elbow height if you have a tape measure — and get desk, chair and monitor heights.