Recent improvements in the capability of statistical or dynamic models to predict climate fluctuations several months in advance may be an opportunity to improve the management of climatic risk in rain-fed agriculture. The aim of this paper is to evaluate the potential benefits that seasonal climate predictions can AU1 bring to farmers in West Africa. The authors have developed an archetypal bioeconomic model of a smallholder farm in Nioro du Rip, a semiarid region of Senegal. The model is used to simulate the decisions of farmers who have access to a priori information on the quality of the next rainy season. First, the potential economic benefits of a perfect rainfall prediction scheme are evaluated, showing how these benefits are affected by forecast accuracy. Then, the potential benefits of several widely used rainfall prediction schemes are evaluated: one group of schemes based on the statistical relationship between rainfall and sea surface temperatures, and one group based on the predictions of coupled ocean-atmosphere models. The results show that forecasting a dryer than average rainy season would be the most useful to Nioro du Rip farmers if they interpret forecasts as deterministic. Indeed, because forecasts are imperfect, predicting a wetter than average rainy season exposes the farmers to a high risk of failure by favoring cash crops such as maize and peanut that are highly vulnerable to drought. On the other hand, the farmers' response to a forecast of a dryer than average rainy season minimizes the climate risk by favoring robust crops such as millet and sorghum, which will tolerate higher rainfall in case the forecast is wrong. When either statistical or dynamic climate models are used for forecasting under the same lead time and the same 31-yr hindcast period (i.e., 1970-2000), similar skill and economic values at farm level are found. When a dryer than average rainy season is predicted, both methods yield an increase of the farmers' income-13.8% fo