The success of influenza vaccination depends largely on the antigenic match between the influenza vaccine strains and the virus strains actually circulating during the season. In the past, this match has proved to be satisfactory in most seasons. In the 1997/1998 season, however, hemagglutination inhibition (HI) assays with ferret antisera indicated a considerable mismatch between the H3N2 vaccine component and the most prevalent epidemic influenza A(H3N2) virus. The results from antigenic analyses using pre- and postvaccination serum samples from volunteers of various ages, including residents of nursing homes who were more than 60 years of age, were in good agreement with the results obtained with ferret antisera. Homologous serum antibody responses to the H3N2 vaccine component as well as the cross-reactivity of the induced antibodies to the epidemic H3N2 strain, declined with increasing age of the vaccinees. As a consequence of these two effects, 84% of the vaccinees over 75 years of age did not develop HI antibody titers ≥ 40 against the major H3N2 virus variant of 1997/1998, suggesting that they were not protected against infection with this virus variant. These findings support the current policy of the World Health Organization (WHO), which is to base worldwide influenza virus surveillance on results predominantly obtained by antigenic analyses of influenza virus isolates with ferret antisera in HI tests. If an antigenic mismatch is observed, the protective efficacy of the vaccine, especially for the elderly, may be insufficient. The observations also support the current policy to include the elderly in serologic efficacy trials. J. Med. Virol. 61:94–99, 2000. © 2000 Wiley-Liss, Inc.