An outbreak of measles among a predominantly unvaccinated and susceptible Amish population in Lebanon County, Pennsylvania, offered the opportunity to test the hypothesis that secondary cases in households are more severe than primary cases because the former have more intense exposure and receive a greater virus inoculum. Of 130 measles cases reported between April and June 1988, 119 (92%) constituted a study of disease severity. Severity was assessed by determining frequency and duration of symptoms, length of any hospitalization, and number of days in bed. In a univariate analysis, fewer secondary cases had conjunctivitis (relative risk [RR], 0.67; 95% confidence interval [CI], 0.48-0.96) and headache (RR, 0.37; CI, 0.15-0.86), but more had earache (RR, 9.69; CI, 1.8-202.9) compared with primary cases. Secondary cases had a shorter mean duration of coryza (4.0 vs. 5.0 days, Student's t test, P = .08). However, a logistic regression model that matched by family and controlled for age and sex indicated that there were no significant differences in measles severity among primary and secondary cases in households.