We conducted a serial cohort study among elderly members of two and three large managed care organizations for the 1996–1997 and 1997–1998 influenza seasons, respectively, to investigate risk factors for death during the influenza season and whether persons at higher risk for dying were also more likely to be vaccinated. Data were obtained from the clinical and administrative databases of the health plans, and multivariable logistic regression was used to assess the impact of specific risk characteristics on the risk of dying and the likelihood of vaccination while controlling for baseline demographic and health characteristics. There were 82,867 and 158,454 subjects in the two cohorts. Subjects with advanced age (≥85 years), medical co-morbidities, hospitalization during the baseline period and hospitalization for pneumonia or influenza during the vaccination period for each study year were more likely to die during the influenza season. However, subjects with advanced age, certain co-morbidities (renal disease or dementia/stroke), hospitalization during the baseline period or hospitalization for pneumonia or influenza during the vaccination period were also less likely to be vaccinated. These disparities in vaccination rates according to risk status may confound interpretations of ecologic studies assessing temporal trends in vaccination rates and influenza-attributable mortality rates and point out the need for more effective strategies to ensure that high-risk persons are immunized.