The clinical efficacy of measles vaccines was investigated in Niakhar, a rural area of Senegal under demographic surveillance in 1987-1990. Three measles vaccines were tested: a standard Schwarz, a high-titer Edmonston-Zagreb, and a high-titer Schwarz. The two high-titer vaccines were administered at 5 months of age and the standard Schwarz vaccine at 10 months. In addition to a formal randomized vaccine trial, data from national campaigns using the standard Schwarz vaccine were also analyzed. Clinical efficacy was estimated after controlling for exposure. In the randomized trial, the estimate of the efficacy of the standard Schwarz vaccine was 97.2% (95% confidence interval (CI) 91.3-98.1). In the 1986-1987 national campaign, the efficacy of the standard Schwarz vaccine was lower: 92.5% (95% CI 88.8-94.6). In the randomized trial, the efficacy of the high-titer vaccines was lower than that of the standard vaccine. High-titer vaccines were not used in national campaigns. Other factors associated with vaccine efficacy were age at exposure, intensity of exposure, and age at vaccination. Controlling for the intensity of exposure did not change the relative ranking of the efficacy of the three vaccination strategies. The theoretical efficacy of the standard measles vaccine for a single unit of exposure was estimated at 98.0%.