BACKGROUND: The probability that a returned traveller with a history of fever has malaria is likely to vary by geographical area, but this has not been quantified in travellers. AIM: To collect data on prevalence of malaria in outpatients returning with a fever or history of fever from malaria-endemic countries, at the point of presentation for a malaria test. DESIGN: Observational retrospective study. Consecutive patients presenting to an unselected 'walk-in' clinic for returned travellers. RESULTS: Of 2867 patients meeting inclusion criteria, 337 (11.8%) had malaria, 89.5% originating in sub-Saharan Africa. Of travellers returning from sub-Saharan Africa excluding South Africa with fever/history of fever, 291/1497 had malaria (19.4%, 95% CI 17-21%). A high proportion was visiting friends and relatives. In those from other areas the proportions were: 16/707 (2.3%, 95% CI 1.5-3.8) from Indian subcontinent/Southeast Asia; 2/143 (1.4%) from Southern America; 4/129 (3.1%) from South Africa; 1/44 (2.3%) from North Africa; and 8/41 (19.5%) from Oceania. Compared to other malaria-endemic regions, African travel gave an adjusted odds ratio of 7.8 (95% CI 5.4-11.2, P < 0.0001). Only 45.1% of malaria cases had a fever (> or =37.5 degrees C) at the time of presentation. Only 3% of all diagnoses of malaria had no history of fever. In 28% of cases parasite count increased in the initial 24 h of antimalarial treatment. CONCLUSION: The likelihood that a patient with fever returning from Africa has malaria is high (around 1 in 5), and is significantly lower from other areas. Absence of fever at presentation does not exclude malaria.