Most cases of malaria in Zimbabwe are diagnosed on the basis of clinical suspicion, without laboratory tests. Of patients treated, between 10 and 30% have malaria parasites on blood slide examination. Can diagnosis be improved by a systematic history? We examined this question in 287 patients treated for malaria in an area of year-round transmission in Zimbabwe. The most common complaints were 'headache' (85.7%), 'bodily weakness' (79.0%) and 'fever/feeling hot' (73.2%). Eighty patients (28%) had malaria parasites on blood smear. Using the blood slide as the standard, we calculated the sensitivity, specificity and positive predictive value of a variety of clinical symptoms and signs. None had a positive predictive value substantially higher than the unknown diagnostic criteria used by health workers (28%). Multivariate analysis showed that 15 different demographic and clinical variables did not significantly predict a positive blood slide result. We conclude that, in this setting, clinical history alone will not improve the diagnosis of malaria.