Background. To assess the effectiveness of the cholera prevention activities of the Peruvian Ministry of Health, we conducted a knowledge, attitudes, and practices (KAP) survey in urban and rural Amazon communities during the cholera epidemic in 1991. Methods. We surveyed heads of 67 urban and 61 rural households to determine diarrhoea rates, sources of cholera prevention information, and knowledge, attitudes, and practices regarding ten cholera prevention measures. Results. Twenty-five per cent of 482 urban and 11% of 454 rural household members had diarrhoea during the first 3-4 months of the epidemic. Exposure to mass media education was greater in urban areas, and education through interpersonal communication was more prevalent in rural villages. Ninety-three per cent of rural and 67% of urban respondents believed they could prevent cholera. The mean numbers of correct responses to ten knowledge questions were 7.8 for urban and 8.2 for rural respondents. Practices lagged behind knowledge and attitudes (mean correct response to ten possible: urban 4.9, rural 4.6). Seventy-five per cent of respondents drank untreated water and 91% ate unwashed produce, both of which were identified as cholera risk factors in a concurrently conducted case-control study. Conclusions. The cholera prevention campaign successfully educated respondents, but did not cause many to adopt preventive behaviours. Direct interpersonal education by community-based personnel may enhance the likelihood of translating education into changes in health behaviours. Knowledge, attitudes, and practices surveys conducted with case-control studies during an epidemic can be an effective method of refining education/control programmes.