A prospective surveillance study of legionnaires disease and an environmental survey of Legionella species were performed simultaneously in a general hospital. During a period of 3 years, 705 patients with pneumonia were screened with a Legionella urinary antigen test, and pneumonia was confirmed by culture and serological tests. Twelve cases of legionnaires disease were identified, none of which were hospital acquired, despite the fact that 60% of hospital water samples were contaminated with Legionella pneumophila at a concentration of more than 10(3) colony-forming units/L. The probable source of infection was identified for only 2 community-acquired cases. The results show that environmental contamination alone is not able to predict the risk of legionnaires disease. If no cases are present, monitoring of hospital water systems is of little significance; clinical surveillance is much more important.