The health effects of bathing in coastal waters is an area of scientific controversy. We conducted the first ever randomised "trial" of an environmental exposure to measure the health effects of this activity. The trial was spread over four summers in four UK resorts and 1216 adults took part. Detailed interviews were used to collect data on potential confounding factors and intensive water quality monitoring was used to provide more precise indices of exposure. 548 people were randomised to bathing, and the exposure included total immersion of the head. Crude rates of gastroenteritis were significantly higher in the exposed group (14.8 per 100) than the unexposed group (9.7 per 100; p = 0.01). Linear trend and multiple logistic regression techniques were used to establish relations between gastroenteritis and microbiological water quality. Of a range of microbiological indicators assayed only faecal streptococci concentration, measured at chest depth, showed a significant dose-response relation with gastroenteritis. Adverse health effects were identified when faecal streptococci concentrations exceeded 32 per 100 mL. This relation was independent of non-water-related predictors of gastroenteritis. We do not suggest that faecal streptococci caused the excess of gastrointestinal symptoms in sea bathers but these microorganisms do seem to be a better indicator of water quality than the traditional coliform counts. Bathing water standards should be revised with these findings in mind.