SUMMARY The development of an amoebal enrichment method, for the recovery of viable Legionella pneumophila from clinical materials is described. The method has been used successfully in five of six cases of Legionnaires' disease in which L pneumophila was isolated.Studies on those L pneumophila isolates led to the discovery that virulent legionellae are attracted to, and attack suitable host amoebae, rather than infection following chance ingestion.Amoebae are probably natural hosts for Legionella pneumophila and similar bacteria,' therefore it is logical to try and use these protozoa to facilitate the isolation of legionellae from clinical specimens. Amoebae could be used in several ways: (i) to increase the number of legionellae in the specimen; (ii) to protect preferentially legionellae from harnmful substances; (iii) to enable isolation of legionellae via micromanipulation. All these strategies have been successfully applied to a sputum sample. Experience suggests that amoebal enrichment followed by acid decontamination and plating on selective media, is a promising method for the isolation of L pneumophila from contaminated clinical material.Amoebal enrichment requires the exposure of viable legionellae to appropriate viable amoebae in a suitable preparation. Legionellae in clinical specimens may be intracellular, and sometimes aflagellate and therefore cannot seek out amoebae. Because in a primary amoebal enrichment an amoeba has to ingest a viable legionella to become infected, it may be several days before the preparation contains numerous infected amoebae. Subsequent enrichments, particularly liquid enrichments where motile legionellae can seek out the amoebae are much quicker, usually 2-3 days at 35°C, a temperature regarded as optimal for the growth of clinical isolates of L pneumophila. To increase the chances of an amoeba coming into contact with a legionella, specimens should be first either liquefied or homogenised, then washed to reduce the concentrations of salt (which can be inhibitory