Influenza epidemics are associated with excess winter mortality. Risk factors for influenza complications and death include chronic illness and living in residential care. In the UK uptake of influenza vaccine among high-risk groups is only 10-40%, partly because of scepticism about vaccine efficacy. We have assessed the efficacy of influenza vaccine in reducing mortality from certified influenza by a case-control study of subjects aged 16 years or older who died between Nov 4, 1989, and Feb 23, 1990, in 36 district health authorities in England. We reviewed general practitioners' records for 315 patients who died of influenza and 777 controls, matched for age, sex, and area of residence, who died a year after the epidemic. Information was collected on demography, the usual place of residence (institutional or noninstitutional), and the existence of chronic illness. Conditional logistic regression analysis for matched case-control studies showed that influenza vaccination reduced mortality by 41% (95% CI 13-60) for all subjects. Further adjustments showed that among subjects who received the vaccine for the first time in 1989, vaccination reduced mortality by 9% (0-59); however, among those who had also been vaccinated previously, mortality was reduced by 75% (31-91). We detected no significant differences in the effect of vaccine between subjects who lived in institutions and in the community (p = 0.16), or between subjects with high-risk medical conditions and those without (p = 0.76). Influenza vaccine is effective in reducing mortality from influenza, and efficacy seems to be greater after repeated annual vaccination than after first administration.