The purpose of this study was to assess the feasibility, acceptability and effect of an in-home water chlorination programme in a rural village. Previous studies at this site showed high levels of faecal coliforms in household water, high diarrhoea rates in children, and enterotoxigenic Escherichia coli and rotaviruses were the most common pathogens isolated from patients. Household water came from a pond and was stored in clay pots. No homes had sanitary facilities. A blind, cross-over trial of treatment of household water with inexpensive hypochlorite by a community health worker was carried out over 18 weeks among 20 families. Water in the clay pots was sampled serially, and symptom surveillance was done by medical students. The programme was generally acceptable to the villagers and no change in water use patterns were apparent. The mean faecal coliform level in the chlorinated water was significantly less than in the placebo treated samples (70 vs 16000 organisms/dl, P less than 0.001). People living in houses receiving placebo treatment had a mean of 11.2 days of diarrhoea per year, and the highest rate of 36.7 was among children less than 2 years old. Diarrhoea rates were not significantly different among the participants while exposed to water treated with hypochlorite. We conclude that a low-cost programme of this type, which utilizes community resources, is logistically feasible, appears to be culturally acceptable in this setting, and can result in a marked reduction in water contamination. The lack of effect on diarrhoea rates suggests that improvement in water quality may affect morbidity only when other variables relating to faecal-oral agent transmission are ameliorated at the same time.