Evidence shows a strong relationship between child mortality and fertility at the aggregate level but the relationship at the individual level is less clear. Data from the 1993 Ghana DHS are used to examine the impact of infant death on a woman's subsequent fertility behaviour. Birth interval analysis, parity progression ratios, and multilevel discrete-time hazard models are used. Child replacement after infant death is found to be taking place in Ghana. On average, birth intervals are shortened by about 15 months if a child dies in the neonatal stage, and by about 11 months for postneonatal death. Progression to the next parity is higher if an infant dies than if it survives; the probability of progression is about 32% higher if a male child dies than if a female dies. A sustained decline in child mortality in Ghana is likely to result in further reduction in fertility.