Anthropometric data for 1295 children were collected during cross-sectional surveys conducted in Kurunegala District, Sri Lanka between March 1987 and March 1988. The relationship between anthropometric status and a range of water, sanitation and hygiene-related exposures was examined. After taking account of socioeconomic and other potential confounding factors, some evidence was found that boiling of water was associated with improved height-for-age. It seems unlikely that boiling of water alone could be responsible for the observed increase in height-for-age. Boiling of water may be acting as a "marker" for a range of hygiene and child-care behaviours. No convincing evidence of any other association was found. In particular, water source was not associated with anthropometric status. A concurrent study conducted in the same population found evidence of a substantial reduction in diarrhoea morbidity associated with the use of improved sources. These findings, taken together, lend support to the hypothesis that interventions for the control of diarrhoea may have no detectable impact on anthropometric status.