To identify risk factors for developing pneumococcal infections, we carried out a case-controlled study on a retrospectively constituted cohort of 3074 clinic patients in a presumed high-risk population. Culture-proved pneumococcal infections were identified in 63 men over a period of 5.5 years, yielding an estimated incidence of 6.3 cases per 1000 person-years. By comparing these patients with 130 uninfected control patients, the relative risk of pneumococcal infections related to various exposures was calculated by logistic regression analysis. Statistically significant independent risk factors (and their relative risks) were as follows: dementia (5.82), seizure disorders (4.38), current cigarette smoking (4.00), congestive heart failure (3.83), cerebrovascular disease (3.82), institutionalization (3.13), and chronic obstructive pulmonary disease (2.38). Risk was increased with age and previous hospitalizations, and, to a nonsignificant degree, by hotel residence (3.93), lung cancer (2.24), previous smoking (2.14), corticosteroid use (1.81), and alcoholism (1.35); but not by diabetes mellitus (0.99), nonlung malignancies (0.93), nonwhite race (0.89), or ischemic heart disease (0.58).