Previous studies of an urban and a rural epidemic in Guinea-Bissau have shown perinatal mortality to be statistically significantly higher among children whose mothers have been exposed to measles during pregnancy. After the epidemic in 1970 in Bandim, a district in the capital of Guinea-Bissau, such children also had a postperinatal childhood mortality risk (7 days to 5 years of age) of 0.229, compared with 0.134 for other children in the community. None of the mothers had developed clinical measles. In a Cox regression analysis adjusting for known background factors, the mortality hazard ratio between the exposed and the controls was found to be 2.0 (95% confidence interval 1.1-3.8). After a small rural measles epidemic in Quinhamel in 1983, the mortality hazard ratio for children of mothers exposed during pregnancy compared with controls was 1.7 (95% confidence interval 0.6-4.6). Exposure to measles or some concomitantly transmitted pathogen during fetal life may contribute to the high childhood mortality found in many developing countries.