Information on the economic burden of malaria in Africa is needed to target interventions efficiently and equitably, and to justify investment in research and control. A standard method of estimation has been to sum the direct costs of expenditure on prevention and treatment, and the indirect costs of productive labour time lost. This paper discusses the many problems in using such data to reflect the burden to society or the potential benefits from control. Studies have generally focussed on febrile illness, overestimating the burden of uncomplicated malaria, but underestimating the costs of severe illness, other debilitating manifestations, and mortality. Many use weak data to calculate indirect costs, which fail to account for seasonal variations, the difference between the average and marginal product of labour, and the ways households and firms 'cope' in response to illness episodes. Perhaps most importantly, the costs of coping mechanisms in response to the risk of disease are excluded, although they may significantly affect productive strategies and economic growth. Future work should be rooted in a sound understanding of the health burden of malaria and the organisation of economic activities, and address the impact on the productive environment, and epidemiological and socio-economic geographical variation.