Accelerating development in South America, with consequent contamination of rivers as the final common pathway of waste, raises concern about adverse effects on the health of riverine populations. We conducted a cross-sectional survey simultaneously in six Latin American nations among people living near a river known to be polluted in each country. Trace metals (arsenic, mercury and lead) were selected as indicators of exposure to industrial effluents. Within each country, we contrasted probability samples from three types of communities: one upstream of point sources of pollution and thus little exposed; one downstream from a site of major development; and one with intermediate exposure. The outcome variables were health status measures elicited by questionnaire and biochemical measures of blood and urine. We examined several possible explanatory and confounding variables, including housing conditions, nutrition, and source of drinking water. The field work was done with triple blinding in that data-gatherers and study subjects were unaware of their group membership and of the study hypotheses, those analysing, specimens and questionnaires were blind to country and community of origin of the material and the investigators reviewed the results in code, committing themselves to conclusions in writing before the codes were broken. Methods were carefully standardized across six countries during training, when pretesting data-gathering instruments and with double coding and extensive accuracy checks of computerized data. There were 900 eligible subjects from 18 communities in six countries. The overall response rate was 92%, the lowest 86%. Results showed an acceptable level of health in all communities and no relationship to exposure.(ABSTRACT TRUNCATED AT 250 WORDS)