We assessed the importance of Streptococcus pneumoniae and immunologic response to 14-valent pneumococcal vaccine in a randomized (saline placebo or vaccine) double-blind pilot study involving 103 patients with chronic obstructive pulmonary disease (COPD). Antibody titers, the flora of the sputum, respiratory infections or pneumonias, and deaths were monitored. The patients' antibody titers before immunization were higher than healthy control subjects. Titers rose normally in those vaccinated but declined more rapidly. Differences between the group receiving placebo and vaccine were not significant at 12 and 24 months. The incidence of pneumonia was high before and after vaccine (47/1,000 vs 41/1,000 patient-years). Nonpneumococcal causes predominated (73 percent of pneumonias; 83.4 percent of lethal pneumonias). Isolates from sputum were predominantly nonvaccine types (50 to 62.5 percent). Twenty-seven patients died; pneumonia occurred in six (one pneumococcal in a vaccinated patient) as a terminal complication of other diseases. Thus, although pneumonia occurred frequently in these patients with COPD and contributed to mortality in 22 percent (six) of the 27 deaths, the predominance of nonpneumococcal causes and the data on antibodies and sputum suggest that pneumococcal vaccine may not be as beneficial for patients with COPD as was hoped. More observations are needed.