In July and August 2007, a giardiasis outbreak affected attendees of a private recreational camp in California. Twenty-six persons had laboratory-confirmed giardiasis; another 24 had giardiasis-like illness with no stool test. A retrospective cohort study determined that showering was associated with illness (adjusted odds ratio 3·1, 95% confidence interval 1·1-9·3). Two days before the outbreak began, the camp had installed a slow-sand water filtration system that included unsterilized sand. Review of historical water-quality data identified substantially elevated total coliform and turbidity levels in sand-filtered spring water used for showering during the suspected exposure period. Unfiltered spring water tested at the same time had acceptable coliform and turbidity levels, implicating the filtration system as the most likely contamination source. To prevent waterborne illness, slow-sand water filtration systems should use sterilized sand, and slow-sand-filtered water should not be used for any purpose where inadvertent ingestion could occur until testing confirms its potability.