The effect of in¯uenza vaccination on the occurrence and severity of in¯uenza virus infection in a population residing in nursing homes for the elderly was studied during an in¯uenza A (H3N2) epidemic in Japan. Of 22 462 individuals living in 301 welfare nursing homes, 10 739 received either one dose (2027 subjects) or two doses (8712 subjects) of inactivated, subunit trivalent in¯uenza vaccine. During the period Nov. 1998 to March 1999, there were 950 cases of in¯uenza infection diagnosed clinically, with virus isolation or serology. There were statistically signi®cantly fewer cases of in¯uenza, hospital admissions due to severe infection and deaths due to in¯uenza in the vaccinated cohort (256 cases, 32 hospital admissions, 1 death) than in the unvaccinated controls (694 cases, 150 hospital admissions, 5 deaths; reduction rates 59.8%, 76.9% and 79.1% respectively). Vaccination was almost equally effective in those who received one dose of vaccine and those who received two doses. No serious adverse reactions to vaccination were recorded. Thus in¯uenza vaccination is safe and effective in this population, and should be an integral part of the routine care of persons aged >65 years residing in nursing homes.