One hundred patients with sporadic, community-acquired, serologically confirmed Legionnaires' disease were matched with control subjects known by the patients (acquaintance controls) and control subjects chosen from among patients with negative serodiagnostic tests for Legionnaires' disease (clinical controls). Each clinical control subject was also matched with an acquaintance control of his own. Legionnaires' disease patients had smoked more cigarettes, consumed more alcohol, and were more likely to have resided near excavation sites than acquaintance or clinical control subjects. Parallel differences between clinical controls and their acquaintances were not seen. Legionnaires' disease patients had traveled away from home for more time during the 2 weeks before onset of illness than had their acquaintances. The difference was of greater magnitude than that between clinical control subjects and their acquaintances. Legionnaires' disease patients were more likely to have resided near construction sites than clinical controls, and there were more construction workers among patients than among clinical control subjects.