Data from the 1993 Ghana Demographic and Health Survey are used to determine whether there is any association between infant mortality and reproductive behavior in Ghana, and to investigate the existence of son preference by examining the link between male child loss and subsequent fertility. Life table analysis of median birth intervals, parity progression ratios and discrete time hazards models are used to study these relationships. The findings of the study indicate that the death of an infant significantly increases a mother's probability of having another birth. Median birth intervals are reduced by 14 months when an index child dies in infancy. The data clearly show that the death of a male child induces a greater reduction in the birth interval (15 months) than that of a female child (13 months). Similarly, parity progression ratios are consistently much higher for women whose children have died, although no obvious gender pattern emerges. This positive association between the death of an infant and the pace of childbearing still persists when confounding variables are taken into account. Further, the probability of having the subsequent birth is 32% greater when the dead child is a male than when the child is a female. However, no differential progression to the next birth is observed by gender in a situation where the index child survives, suggesting a higher tendency for mothers in Ghana to replace dead males. This finding implies some degree of son preference and may stem from a genuine desire to have male children in the family to carry on lineage duties.