Although the protective efficacy of pneumococcal polysaccharide vaccine has been demonstrated in randomized trials in young African gold miners, there has been controversy about its efficacy in older Americans at risk for serious pneumococcal infections. To assess the vaccine's protective efficacy against invasive pneumococcal infections, we conducted a hospital-based case-control study of the efficacy of pneumococcal vaccine in adults with a condition recognized to be an indication for receiving the vaccine. From 1984 to 1990, adults in whom Streptococcus pneumoniae was isolated from any normally sterile site were identified by prospective surveillance in the microbiology laboratories of 11 large hospitals; those with an indication for pneumococcal vaccine were enrolled as case patients. For each case patient, one control was matched according to age, underlying illness, and site of hospitalization. We contacted all providers of medical care to ascertain each subject's history of immunization with pneumococcal vaccine. Isolates of S. pneumoniae were serotyped by an investigator unaware of the subject's vaccination history. Thirteen percent of the 1,054 case patients and 20 percent of the 1,054 matched controls had received pneumococcal vaccine (P less than 0.001). When vaccine was given in either its 14-valent or its 23-valent form, its aggregate protective efficacy (calculated as a percentage: 1 minus the odds ratio of having been vaccinated times 100) against infections caused by the serotypes represented in the vaccine was 56 percent (95 percent confidence interval, 42 percent to 67 percent; P less than 0.00001) for all 983 patients infected with a serotype represented in the vaccine, 61 percent for a subgroup of 808 immunocompetent patients (95 percent confidence interval, 47 percent to 72 percent; P less than 0.00001), and 21 percent for a subgroup of 175 immunocompromised patients (95 percent confidence interval, -55 percent to 60 percent; P = 0.48). The vaccine was not efficacious against infections caused by serotypes not represented in the vaccine (protective efficacy, -73 percent; 95 percent confidence interval, -263 percent to 18 percent; P = 0.15). Polyvalent pneumococcal vaccine is efficacious in preventing invasive pneumococcal infections in immunocompetent patients with indications for its administration. This vaccine should be used more widely.