Although all-cause mortality has been used as an indicator of the health status of childhood populations, such data are sparse for most rural areas of sub-Saharan Africa, particularly community-based estimates of infant mortality rates. The longitudinal follow-up of more than 1,500 children enrolled at birth into the Asembo Bay Cohort Project (ABCP) in western Kenya between 1992 and 1996 has provided a fixed birth cohort for estimating all-cause mortality over the first 5 yr of life. We surveyed mothers and guardians of cohort children in early 1999 to determine survival status. A total of 1,260 households were surveyed to determine the survival status of 1,556 live births (99.2% of original cohort, n = 1,570). Most mothers (66%) still resided but 27.5% had migrated, and 5.5% had died. In early 1999, the overall cumulative incidence of all-cause mortality for the entire 1992-1996 birth cohort was 26.5% (95% confidence interval, 24.1-28.9%). Neonatal and infant mortality were 32 and 176 per 1,000 live births, respectively. These community-based estimates of mortality in the ABCP area are substantially higher than for Kenya overall (nationally, infant mortality is 75 per 1,000 live births). The results provide a baseline description of all-cause mortality among children in an area with intense Plasmodium falciparum transmission and will be useful in future efforts to monitor changes in death rates attributable to control programs for specific diseases (e.g., malaria and HIV/AIDS) in Africa.