OBJECTIVE: To describe the epidemiology of malaria in Guinea-Bissau, in view of the fact that more funds are available now for malaria control in the country. METHODS: From May 2003 to May 2004, surveillance for malaria was conducted among children less than 5 years of age at three health centres covering the study area of the Bandim Health Project (BHP) and at the outpatient clinic of the national hospital in Bissau. Cross-sectional surveys were conducted in the community in different malaria seasons. RESULTS: Malaria was overdiagnosed in both health centres and hospital. Sixty-four per cent of the children who presented at a health centre were clinically diagnosed with malaria, but only 13% of outpatient children who tested for malaria had malaria parasitaemia. Only 44% (963/2193) of children admitted to hospital with a diagnosis of malaria had parasitaemia. The proportion of positive cases increased with age. Among hospitalized children with malaria parasitaemia, those less than 2 years old were more likely to have moderate anaemia (RR = 1.27; 95% CI: 1.02-1.56) (P = 0.03) or severe anaemia (RR = 1.67; 95% CI: 1.25-2.24) (P = 0.0005) than older children. The prevalence of malaria parasitaemia in the community was low (3%, 53/1926). CONCLUSION: In Bissau, the prevalence of malaria parasitaemia in the community is now low and malaria is over-diagnosed in health facilities. Laboratory support will be essential to avoid unnecessary use of the artemisinin combination therapy which is now being introduced as first-line treatment in Bissau with support from the Global Fund.