From January to April, 1990, 695 measles cases were reported to the Puerto Rico Health Department; there were 12 measles-associated deaths (case fatality ratio, 17/1000), more than in any year since 1967. We conducted a case-control study of risk factors for severe measles. We identified 16 children (ages 5 to 34 months) with severe measles and selected children with nonsevere measles as controls (39 hospitalized and 38 nonhospitalized). Controls were frequency matched to severe measles cases by region of residence. One case and two controls had been vaccinated. An underlying illness was present in 50% of cases and 16% of nonhospitalized controls (Mantel-Haenszel weighted odds ratio 5.3; 95% confidence interval 1.4, 20.2). In a multivariate analysis cases were significantly more likely than hospitalized controls to be from families with an annual income of < $5000 (odds ratio (OR), 26.9), to have a mother without a high school degree (OR 11.1), to be anemic (hemoglobin < 10 g/dl) (OR 15.9) and have an underlying illness (OR 18.3). During measles outbreaks preventing severe illness requires aggressive control measures and strategies to increase vaccine coverage of children with underlying illnesses and of low socioeconomic status.