SOUTHERN Africa is subject to recurrent droughts which cause severe food shortages. There is considerable evidence(1) that El Nino(2) warm events in the Pacific Ocean are linked to below-average rainfall in southern Africa, and the 1991-92 El Nino event was accompanied by the worst drought in southern Africa this century, affecting nearly 100 million people. But although models can predict El Nino events a year in advance(3-6), the drought was not anticipated, increasing relief costs. Here we present data showing a strong correlation between an El Nino index and both rainfall and maize yield in Zimbabwe. Surprisingly, the correlation with maize yield is stronger than that with rainfall, with more than 60% of the variance in yield accounted for by sea surface temperatures in the eastern equatorial Pacific Ocean-half-way around the world. We also show that model predictions of the El Nino index provide accurate forecasts of maize yield in Zimbabwe, with lead times of up to a year. As maize is the most important food crop for the ten-nation Southern African Development Community region(7), we suggest that this approach could provide an effective early-warning system for southern African drought-induced famines.