This study compares how effectively the ponderal index and the body mass index adjust birthweight for length at different gestations, and derives an improved index suitable for all gestations. The study was a cross-sectional survey, in a London teaching hospital, using a total of 999 neonates of 33 weeks gestation or later. Main outcome measures were the ponderal index (birthweight/length3), body mass index (birthweight/length2), and Benn index (birthweight/length(n)), where the length power n varies with gestation and is estimated by log-log regression. Results showed that up to 39 weeks gestation, the ponderal index is uncorrelated with length and so is a good index of birthweight for length. Past 39 weeks gestation, the ponderal index is negatively correlated with length, while the body mass index is uncorrelated, so that the body mass index is better. Neither index is optimal at all gestations. Deriving the Benn index (birthweight/length(n)) for each week of gestation, choosing n to make the index uncorrelated with length, shows that n falls steadily and very significantly (p < 0.0001) with increasing gestation. This in turn means that predicted birthweight for length depends on gestation: for a neonate 48 cm long, predicted birthweight varies from 2485 g at 34 weeks to 3030 g at 43 weeks, a 20% range. However, for a 54 cm long infant, predicted birthweight is the same at all gestations. A Benn index where the value of n changes linearly with gestation is described. We conclude that the ponderal index is not appropriate for measuring intra-uterine malnutrition, as it fails to adjust for length at all gestations. No other index of birthweight/length(n) with constant n is any better, as different gestations require different indices. Birthweight predicted from an infant's length depends on the infant's gestation. If, as Barker proposes, thinness at birth assessed by birthweight for length is used to predict later health status, more account needs to be taken of the complex relationship between birthweight, length and gestation.