The epidemic of cholera that began in Peru in January, 1991, marked the first such epidemic in South America this century. Subsequently, over 533 000 cases and 4700 deaths have been reported from nineteen countries in that hemisphere. We investigated the epidemic in Trujillo, the second largest city in Peru. Trujillo's water supply was unchlorinated and water contamination was common. Suspect cholera cases were defined as persons presenting to a health facility with acute diarrhoea between Feb 1, and March 31, 1991. We studied a cohort of 150 patients who had been admitted to hospital and conducted a matched case-control study with 46 cases and 65 symptom-free and serologically uninfected controls; we also carried out a water quality study. By March 31, 1991, 16 400 cases of suspected cholera (attack rate 2·6%), 6673 hospital admissions, and 71 deaths (case-fatality rate 0·4%) had been reported in the province of Trujillo. 79% of stool cultures of patients with diarrhoea presenting to a single hospital yielded Vibrio cholerae 01. In the case-control study, drinking unboiled water (odds ratio [OR] 3·1, 95% confidence interval [Cl] 1·3-7·3), drinking water from a household water storage container in which hands had been introduced into the water (4·2, 1·2-14·9), and going to a fiesta (social event) (3·6, 1·1-11·1) were associated with illness. The water quality study showed progressive contamination during distribution and storage in the home: faecal coliform counts were highest in water from household storage containers and lowest in city well water. V cholerae 01, biotype El Tor, serotype Inaba, was isolated from three city water samples. Cholera control measures in Trujillo should focus on treatment of water and prevention of contamination during distribution and in the home. Trujillo's water and sanitation problems are common in South America; similar control measures are needed throughout the continent to prevent spread of epidemic cholera.