Pneumococcal vaccine effectiveness was assessed in a randomized trial among 1,686 old people (mean age: 74, standard deviation: 4 years) living in 24 geriatric hospitals and 26 homes for the aged in our district; 937 were vaccinated with Merck-Sharp and Dohme pneumococcal vaccine (14 serotypes). The 749 others composed the reference group. This study was performed during 2 years, since December 1980. Both groups were randomized after a two-criteria stratification: by clinical risk assessed before the study, and by type of homes for the aged. Forty pneumonias were diagnosed, with 13 proved pneumococcal etiology. The incidence of pneumonia was significantly reduced in the vaccinated group (p less than 10(-4) but the mortality rate was not modified. We concluded in favor of the effectiveness of pneumococcal vaccine: etiological fraction 77.1% (51.2%-89.3% confidence limits, 95% risk) in the population we studied. The incidence of pneumococcal-proved pneumonia was not significantly reduced.