The authors estimated the incidence rates of cholera and death between 1985 and 1988 for 32,642 age- and sex-eligible persons who did not participate in a randomized, placebo-controlled field trial of killed oral cholera vaccines in rural Bangladesh. As compared with 20,744 placebo recipients, the relative risk of cholera for all nonparticipants, adjusted for potentially confounding demographic variables, was 1.20 (95% confidence interval (CI) 1.03-1.41); this adjusted relative risk reflected elevated adjusted relative risks in nonparticipants who were medically ineligible (RR = 1.65; 95% CI 1.22-2.22) or refused to participate (RR = 1.19; 95% CI 1.01-1.41), but not in persons absent at the time of vaccination (RR = 1.00; 95% CI 0.78-1.28). The adjusted relative risk of death was also elevated in nonparticipants as compared with placebo recipients (RR = 1.28; 95% CI 1.10-1.48), with the same pattern of adjusted relative risks for different categories of nonparticipants: for ineligible subjects, 2.64 (95% CI 2.12-3.29); for refusers, 1.20 (95% CI 1.02-1.41); and for absentees, 0.95 (95% CI 0.75-1.22). The authors concluded that nonparticipation was associated with clinically cogent adverse health outcomes, but that the magnitude of these associations varied according to the reason for nonparticipation. These findings underscore the caution required in assessing vaccine efficacy with controls who are not vaccinated because of choices made by patients or vaccinators.