The purpose of this study was to investigate the various factors affecting the incidence of adverse reactions after influenza immunization with either a subunit or an inactivated whole-virus vaccine. A total of 1,959 nursing home residents in Geneva (mean age: 84 +/- 8 years, sex ratio: 3F/1M) were randomly allocated to the two vaccine groups. Results showed that the incidence of local adverse reactions could be reduced by using the subunit vaccine injected into the buttock or the internal aspect of the thigh, in preference to the deltoid muscle. Moreover, this study demonstrated that intramuscular injection lowers the risk of local reactions but tends to increase the incidence of general reactions. Implementation of these recommendations should reduce the frequency of side effects due to influenza vaccination and, consequently, raise the rate of acceptance and the immunization coverage of the populations exposed to risk.