It has been suggested that measles infection mainly kills frail children who are likely to die anyhow of other infections. If that were true, the proportion of frail children should increase after the introduction of measles vaccination and post-measles mortality compared with mortality in uninfected children should increase when the case fatality declines and frail children are no longer dying of measles. The latter deduction was investigated in Niakhar, Senegal, where the measles case fatality has declined markedly. Measles has been studied in Niakhar during 12 years from 1983 to 1994. We compared long-term mortality after measles infection in periods with both high and low case fatality. The acute measles case fatality rate (CFR) declined from 6.5% in 1983-1986 to 1.5% in 1987-1994, an age-adjusted decline of 66% (RR=0.34 (0.19-0.58)). Between 1983-1986 and 1987-1994, mortality in the first year after measles infection declined by 35% (RR=0.65 (0.37-1.16)), the pattern being the same in the second and third year after infection (RR=0.63 (0.33-1.21)). This reduction could not be related to introduction of immunization, treatment of measles with Vitamin A, or prophylactic use of antibiotics. Controlling for age, immunization, and season, the decline in post-measles mortality was similar to the fall in non-measles-related mortality between the two periods (mortality rate ratio=0.72 (0.64-0.80)). Since the mortality decline in survivors of measles was as large as the decline in mortality among uninfected children, reduction in acute measles mortality did not lead to accumulation of frail children. We doubt measles infection ever eliminated mainly weak children; it always killed a broad spectrum of children, most of whom were "fit to survive". Hence, it seems unlikely that measles vaccination has contributed to the survival of more frail children.