We have produced maps of Plasmodium falciparum malaria transmission in West and Central Africa using the Mapping Malaria Risk in Africa (MARA) database comprising all malaria prevalence surveys in these regions that could be geolocated. The 1846 malaria surveys analysed were carried out during different seasons, and were reported using different age groupings of the human population. To allow comparison between these, we used the Garki malaria transmission model to convert the malaria prevalence data at each of the 976 locations sampled to a single estimate of transmission intensity E, making use of a seasonality model based on Normalized Difference Vegetation Index (NDVI), temperature and rainfall data. We fitted a Bayesian geostatistical model to E using further environmental covariates and applied Bayesian kriging to obtain smooth maps of E and hence of age-specific prevalence. The product is the first detailed empirical map of variations in malaria transmission intensity that includes Central Africa. It has been validated by expert opinion and in general confirms known patterns of malaria transmission, providing a baseline against which interventions such as insecticide-treated nets programmes and trends in drug resistance can be evaluated. There is considerable geographical variation in the precision of the model estimates and, in some parts of West Africa, the predictions differ substantially from those of other risk maps. The consequent uncertainties indicate zones where further survey data are needed most urgently. Malaria risk maps based on compilations of heterogeneous survey data are highly sensitive to the analytical methodology.