Childhood diarrhoea can be transmitted within the household and community environments. This study examined the effect of partial latrine coverage on diarrhoeal morbidity at the household and community levels. Weekly morbidity data on 272 children, aged less than five years, were collected for 45 weeks from October 1996 to August 1997 in two neighbouring semi-arid communities in rural Zimbabwe. The communities were similar except for sanitation. In Community A, 62% of the children lived in households with a latrine, and in Community B, there was no sanitation. Diarrhoeal morbidity was 68% lower in Community A than in Community B. In addition, the children from the households in Community A without a latrine had lower diarrhoeal morbidity than the children in Community B. The inter-community difference in diarrhoeal morbidity was likely to be due to the latrines and associated hygiene education interrupting intra- and inter-household transmission of diarrhoea. The latrines reduced faecal contamination of the community environment, and as diarrhoeal morbidity fell, contact between infectives and susceptibles within the community would also have fallen. The neighbourhoods in Community A where the adjacent households did not have sanitation had higher diarrhoeal morbidity than in the neighbourhoods where one household did have sanitation. The results of the study suggest that sanitation programmes, which share building costs between the state and individuals, i.e. households, and hence, do not achieve complete coverage in a community, benefit the whole community, because the households that cannot afford a latrine experience lower diarrhoeal morbidity due to the safer community environment.