Sixty-nine laboratory-documented cases of Legionnaires' disease occurred in Vermont between 1 May and 31 December 1977. Clinical manifestations were similar to those in the 1976 Philadelphia epidemic. Case-control studies suggested that Legionnaires' disease patients were more likely to present with headache or diarrhea than were patients with pneumonia of presumed nonbacterial cause. The case-fatality ratio for patients treated with erythromycin was 4%, compared with 17% in patients not treated with erythromycin. Thirteen patients had been hospitalized throughout the 10 days preceding onset of illness, equaling the maximal known incubation period. This suggests either acquisition or reactivation of infection in the hospital. However, even during the week of peak disease activity, cases occurred in patients with no recent hospital contact. The only community factor possibly associated with acquisition was home air conditioning. The prevalence of seroreactivity to the Legionnaires' disease bacterium in various community populations was as high as 26%, suggesting a possible endemic area.