This thesis investigates the distribution of legionellae in water systems in the Royal Liverpool University Hospital (RLUH) and examines some of the factors that affect colonisation by these organisms. The effect of persistent contamination of the domestic water system on immunocompromised patients was monitored, and the envirorunental control of legionellae by various methods was assessed. A fluorescent monoclonal antibody (DFA) was evaluated for its ability to detect L. pneumophila in domestic and cooling water, and was highly sensitive and specific for this purpose. DFA detected non-culturable L. pneumophila in the cold water system (CWS) that were not recovered following heat shock procedures. Legionellae were not isolated from air conditioning humidifiers, and were rarely detected in cooling towers despite treatment with inadequate concentrations of biocide. A high pH assisted in preventing legionella colonisation. Calorifier sediment contained legionellae and high levels of insoluble copper oxides. Culture media and a low pH, released Cuions from sediment which were markedly inhibitory to legionellae. Low concentrations of Cuions were detected in domestic hot water. At temperatures below 60°C legionellae were detected in the hot water supply to the wards, and calorifiers were regularly re-seeded by legionellae returning from contaminated peripheral parts of the system. Legionellae were not detected in the HWS when 60°C was achieved. L. pneumophila sgps 6, 12 and L. bozemanii predominated in domestic water. L. pneumophila sgp 1 was detected on one occasion only in a cold water storage tank and a calorifier, and did not colonise any of the water systems. L. pneumophila sgps 6 and 12 were isolated from three nosocomial cases of Legionnaires' disease. Endemic legionellae prepared as yolk sac antigens, detected significant titres of legionella antibodies (~ 1 :64) in samples from six subjects which did not react ( < 1: 16) with the PHLS L. pneumophUa sgp 1 yolk sac antigen. Most raised titres were to L. pneumophila sgp 12, and the highest titre in heterologous responses identified the infecting serogroup of L. pneumophila. Routine culture of respiratory samples from susceptible patients. detected only one undiagnosed case of Legionnaires' disease. Legionellae were not detected in water from showers that were regularly flushed or irradiated with UV light. Re-colonisation of showers by legionellae was closely associated with the reappearance of amoebae. A trace heating element was effective at maintaining dead-legs at 50°C (± 1.5) and reduced legionellae in these sites. Legionellae proliferated where pipes and heating element were not adequately insulated. Re-circulating the HWS through dead-legs eradicated legionellae from this site but resulted in heavy colonisation of adjacent mixer valves. Automatic drain valves failed to prevent legionellae from colonising shower hoses and mixer valves, and hyperchlorination of shower hoses and water strainers had only a short term effect. Showers heated electrically at point of use were not colonised by legionellae entering in the CWS, or by wild strains of legionellae introduced with calorifier sediment. This appeared to be due to rapid throughput of water, extensive use of copper, and pasteurisation of calorifier contents following discharge of heat from the heating elements, after the shower ceased operating.