A study was conducted to characterize the risk of Plasmodium vivax infection in the Lacandon forest, southern Mexico. Blood samples and questionnaire data were collected in 1992. Malaria cases (n = 137) were identified by the presence of symptoms and a positive thick blood smear. The control group included individuals with negative antibody titres and no history of malaria (n = 4994). From 7628 individuals studied, 1006 had anti-P. vivax antibodies. Seroprevalence increased with age. Risk factors associated with infection included: place of birth outside the village of residence (odds ratio, OR 11·67; 95% CI 5·21–26·11); no use of medical services (OR 4·69, 95% CI 3·01–7·29), never using bed-nets (OR 3·98, 95% CI 1·23–12·86) and poor knowledge of malaria transmission, prevention and treatment (OR 2·30, 95% CI 1·30–4·07). Health education represents the best recommendation for controlling the disease in the area.