This paper reports on 164 cases of kala-azar observed in the Baringo District of Kenya between February 1981 and February 1983. All were confirmed serologically by enzyme-linked immunosorbent assay (ELISA) and all but 20 by parasitological examination as well. Following the standard treatment with a 30 day course of sodium stibogluconate (Pentostam) two non-responders and four relapses were observed. Children between 2 and 15 years old were found to be the most affected age group; male patients predominated slightly at 57%. All cases occurred in the semi-arid and arid parts of the district below 1500 m, where pastoralism predominates. Besides scattered cases, certain kala-azar foci could be identified. Two of these--Endao with 49 households, 228 inhabitants and 13 cases of kala-azar, and Koriema with 22 households, 93 inhabitants and 11 cases--were subject to a house to house survey. People were examined physically, their weight and height recorded and fingerprick blood collected on blotting paper for later serological testing. Each household was mapped and the relevant environmental factors recorded. A positive correlation could be demonstrated between kala-azar cases and the vicinity of their homesteads to seasonal rivers and also between kala-azar cases and people living in timber houses, rather than mud and wattle houses. Eroded termite hills were not found to be of epidemiological importance. No satisfying explanation could be found for the striking temporal and local clustering of cases. The homestead was identified as an important site of transmission with optimum conditions for transmission occurring during supper in the evening. Based on spleen rates, Endao was classified as hyperendemic for malaria and Koriema as mesoendemic. Diagnostic ELISA values above 0.2 were observed in all cases of active kala-azar. However, ELISA values above 0.04, taken as the borderline non-specific reaction, could be found in about half of the study areas population. Therefore we conclude that asymptomatic infection must be common. Observations demonstrated that spontaneous recovery may follow clinical illness and visceralization of the parasite. Comparison of parasitological and serological data suggest that this may be expected in more than 15% of cases.