A number of formulae have been suggested for estimating the surface area (SA) of a human body from measurements of height H and weight W. Most of these are of the same functional form, namely InSA = a(0) + a1lnH + a2lnW in logarithmic terms, but have quite different values of the coefficients. We show that they are all essentially equivalent in view of the strong linear relation between lnH and lnW. The formula due to Gehan and George, in which a(0) = -3.751, a1 = 0.422 and a2 = 0.515 if height is measured in cm, weight in kg and surface area in m2, is based on a sample of 401 surface area measurements and has coefficients estimated by least squares. It should be the medical standard. Moreover, by extending their analysis, it is possible to derive standard errors of surface area estimates and to construct confidence and prediction intervals. Unfortunately, in clinical practice a relation based on just nine subjects, and with coefficients determined in an ad hoc way, is still in common use.