Risk factors such as maternal age, parity, previous siblings' death, inbreeding of parents, birth weight, birth length were examined in a population-based prospective study in four population groups at different levels of urbanization in and round Lahore, Pakistan. From September 1984 to March 1995, 2967 full-term, single born infants were followed from the 5th month of gestation to 12 months of age. Logistic regression analysis showed a significant relative risk (RR) of infant death associated with parents' consanguinity (RR = 1.8), birth weight (RR = 1.8) and elder siblings' death (RR = 1.7). The risk attributed to these factors was 28, 17 and 25%, respectively. The number of lethal equivalents per gamete is about one. The B/A ratio 10.36 suggests that the genetic load is likely to be mutational. In countries like Pakistan, where consanguinity is favourably practiced, a substantial proportion of infant deaths may be prevented by cessation of such marriages. The implications of this finding for the Pakistani community are discussed.