BACKGROUND: From June to November 2005, 18 cases of community-acquired Legionnaires' disease (LD) were reported in Rapid City South Dakota. We conducted epidemiologic and environmental investigations to identify the source of the outbreak. METHODS: We conducted a case-control study that included the first 13 cases and 52 controls randomly selected from emergency department records and matched on underlying illness. We collected information about activities of case-patients and controls during the 14 days before symptom onset. Environmental samples (n = 291) were cultured for Legionella. Clinical and environmental isolates were compared using monoclonal antibody subtyping and sequence based typing (SBT). RESULTS: Case-patients were significantly more likely than controls to have passed through several city areas that contained or were adjacent to areas with cooling towers positive for Legionella. Six of 11 case-patients (matched odds ratio (mOR) 32.7, 95% CI 4.7-infinity) reported eating in Restaurant A versus 0 controls. Legionella pneumophila serogroup 1 was isolated from four clinical specimens: 3 were Benidorm type strains and 1 was a Denver type strain. Legionella were identified from several environmental sites including 24 (56%) of 43 cooling towers tested, but only one site, a small decorative fountain in Restaurant A, contained Benidorm, the outbreak strain. Clinical and environmental Benidorm isolates had identical SBT patterns. CONCLUSION: This is the first time that small fountain without obvious aerosol-generating capability has been implicated as the source of a LD outbreak. Removal of the fountain halted transmission.