During the 1984–1985 influenza season, outbreaks of influenza A (H3N2) occurred in three Connecticut nursing homes. Influenza vaccination rates were 67% (96 out of 144), 35% (30 out of 85) and 69% (332 out of 483), respectively. The relative risk of illness for vaccinated compared to unvaccinated residents was 1.8 (95% confidence interval, 0.6, 5.9), 1.6 (95% confidence interval, 0.8, 3.0) and 1.1 (95% confidence interval, 0.8, 1.7) for each of the three nursing homes, respectively. In the third outbreak, 22 vaccinated residents without clinical illness had a geometric mean titer of hemagglutination-inhibition (HI) antibody of 20. Although low, this titer was significantly higher than that of nine unvaccinated residents without clinical illness (12, p<.05); only three (14%) vaccinated residents had HI titers of 340. These results suggest that levels of HI antibody in vaccinated residents were not protective at the time of the third outbreak, four to five months after vaccination. In general, the study of vaccine effectiveness in nursing homes is limited by sample size and statistical power. Despite these limits, the retrospective investigation of influenza outbreaks in nursing homes is often the only practical way to evaluate influenza vaccine effectiveness in the elderly on a yearly basis.