Global climatic change is expected to increase the incidence of vector-borne diseases, especially malaria. This study assessed the contribution of climate to a malaria epidemic in Rwanda, focusing on the catchment area of one health centre where diagnosis was consistent and non-climatic variables well monitored. In late 1987 malaria incidence in the area increased by 337% over the 3 previous years. The increase was greatest in groups with little acquired immunity--children under 2 years (564%) and people in high-altitude areas (501%). Case-fatality rose significantly (relative risk = 4.85, p < 0.001). 1987 also saw record high temperatures and rainfall. An autoregressive equation including lagged effects of these two variables explained 80% of the variance in monthly malaria incidence. Temperature (especially mean minimum) predicted incidence best at higher altitudes where malaria had increased most. Empirically derived relations were consistent with the estimated generation time of the disease and with the known sensitivity of the plasmodium parasite to temperature. The patterns of climatic warming between day and night and among seasons will be critical to the effect on malaria. These findings are most relevant to regions near the altitude or latitude limits of the disease, where several epidemics have lately been reported.