Data
What the published data actually says
One large meta-analysis, measurements taken by health professionals rather than self-reported. The numbers are lower and the distribution is narrower than almost anyone expects.
The reference most people should be reading is Veale et al. (2015), published in BJU International. It pooled 17 studies covering 15,521 men, and its inclusion criterion is the reason it matters: every measurement was taken by a health professional using a standard method, not reported by the participant.
That single criterion is what separates it from essentially everything else circulating online.
The figures
| Measurement | Mean |
|---|---|
| Flaccid length | 9.16 cm (3.61 in) |
| Flaccid stretched length | 13.24 cm (5.21 in) |
| Erect length | 13.12 cm (5.17 in) |
| Flaccid circumference | 9.31 cm (3.66 in) |
| Erect circumference | 11.66 cm (4.59 in) |
Two things about this table surprise people every time.
The first is that erect length and stretched flaccid length are almost identical. That is not a coincidence - it is why stretched flaccid is used clinically as a proxy, since it is far easier to standardise in a consulting room.
The second is simply that the erect figure is around 13 cm. The number most people carry in their head is larger, and the reason for that gap is well documented.
The distribution is narrow
The paper's central contribution is a set of nomograms - percentile curves rather than single averages - and their shape is the useful part.
The practical consequence: the range containing most people is only about three centimetres wide. Small absolute differences move you a long way through the percentiles, and large differences are genuinely uncommon in both directions.
It also means the difference between a careful measurement and a sloppy one - which is larger than it sounds - can be worth ten or more percentile points on its own. Measurement error is not a small effect against a distribution this tight.
What the paper does not claim
It does not report ethnic differences - the authors state the data were insufficient to draw conclusions, which has not stopped the paper being cited for exactly that.
It does not establish a relationship between size and height, foot size, or anything else in the folklore. Correlations reported elsewhere are weak enough to be useless for prediction at an individual level.
And it does not say anything about anybody's satisfaction with anything. It is a set of measurements.
Reading your own number against it
Measure the way the method describes, take the median of three, and compare to the nomograms rather than to the mean. A percentile is more informative than a distance from average, particularly on a distribution this narrow.
Then hold it loosely. A percentile is a fact about a distribution of measurements. It is not an assessment, and if an assessment is what you were after, that is a different kind of judgement entirely - the sort of thing Cock Rate produces - and it is not derived from a length in centimetres at all.