A community-based intervention was developed through direct participation of the target population in assessment and iterative trials to improve hygiene practices and to reduce childhood diarrhoea in lowland rural Bangladesh. A total of 185 (98%) households with children ages 0-18 months in five contiguous villages were targeted for the interventions. A comparison site was selected for a detailed observational study and for use as a control for the intervention. About 97% of all households with children ages 0-18 months were enrolled for study at the control site. Children in this age group were targeted because at this developmental stage they were most vulnerable to diarrhoeal morbidity and malnutrition (related to unhygienic practices). The intervention was implemented with the assistance of village leaders through a "Clean Life" campaign by local project workers and volunteer mothers who were chosen from the target households. The intervention activities started in January 1986 and lasted for 7 months. Higher adoption rates of the intervention were associated with better cleanliness status, which was related to lower diarrhoea and malnutrition rates in the intervention site. The results of between-site longitudinal analyses showed that after the intervention, the intervention site had substantially higher cleanliness scores, lower diarrhoeal morbidity, and better growth status compared to those of the control site, with differences increasing over time. The findings suggest that this type of community-based intervention can be very beneficial in modifying hygiene behaviours and lowering childhood diarrhoea and malnutrition.