In this paper we assess the importance of heterogeneity and selective fertility in altering estimates and interpretations of the determinants of the human capital of children. We set out a sequential model of human capital investments in children incorporating endogenous fertility and heterogeneity in human capital endowments to illustrate the fertility selection problem and issues of identification. Empirical results based on parametric and semi-parametric estimates of selectivity models applied to data on birthweight and schooling in Malaysia indicate that the hypothesis of no fertility selection is strongly rejected, with mothers having higher birthweight children tending to have substantially lower birth probabilities (negative birth selectivity). As a consequence, the positive association between mother's schooling and birthweight is substantially underestimated and the positive effects of delaying childbearing overestimated when birth selectivity is not taken into account. The schooling results indicate strong rejection of the "efficient schooling" model, in which schooling is allocated efficiently across children, but only when the selectivity of fertility is taken into account.