As an antecedent of birthweight and in its own right, gestational age is an important proximate determinant of infant mortality. Recent analyses using mixture models of birthweight distributions suggest that substantial heterogeneity occurs within a birth cohort even when controlling for sex and ethnicity. This article extends the mixture model analysis to gestational age. The results indicate that, like birthweight, human gestational age distributions are heterogeneous, consisting of two, or perhaps more, subpopulations with separate means and variances. The possibility that birthweight and gestational age both identify the same underlying subpopulations cannot be rejected. Statistical analyses of the sex and ethnic differences indicate that, like birthweight, gestational age distributions vary significantly between the sexes and among ethnic groups. However, the pattern, and even the direction, of the variation often differs between the two indicators of birth outcome. The results suggest that a multivariate mixture model that combines birthweight and gestational age might be a useful extension of the univariate mixture models. Am. J. Hum. Biol. 12:181–191, 2000. © 2000 Wiley‐Liss, Inc.