Brief biweekly home visits, made as part of a cohort study of diarrhea in young children under age 5 years that was carried out in Salvador, Brazil, in 1998-1999, were used as a low-cost way to collect structured observation data on domestic hygiene behavior. Field-workers were trained to check a list of 23 forms of hygienic or unhygienic behavior by the child or the child's caretaker, if any behaviors were seen during the visit. Children were grouped according to whether mainly unhygienic behavior or mainly hygienic behavior had been recorded. This permitted study of the determinants of hygiene behavior and of its role in the transmission or prevention of diarrheal disease. Observations were recorded on roughly one visit in 20. Households with adequate excreta disposal were significantly more likely to be in the "mainly hygienic" group. The prevalence of diarrhea among children for whom mainly unhygienic behavior was recorded was 2.2 times that among children in the "mainly hygienic" group. The relative risk for prevalence was 2.2 (95% confidence interval: 1.7, 2.8). The relative risk fell to 1.9 (95% confidence interval: 1.5, 2.5) after data were controlled for confounding, but the difference was still highly significant.