Rainfed agriculture is increasingly uncertain in the Sahel leading to low agricultural productivity and farmers' discouragement. Most sahelian farmers cape with agricultural risk developing strategies through sale and purchase of grains and small ruminants, which a called recourses in microeconomics. Our objective in this study is to propose a method of evaluating the farm optimal plans under various scenarios of climate variability for farmers who possess bath rainfed and small irrigation plots. The hypothesis is usual recourses are insufficient but that even very small irrigation plots have a much more positive impact, even on the production of the rainfed crops. We use stochastic programming with recourse to improve the analysis of farm management facing risk, which is an improvement compared to the common deterministic models. MonteCarlo simulation and Bender decomposition process help improve the mathematics behind fanners ' decision tree. The model was applied to a typical farm in the north of Burkina where extensive researches have been carried out on climate, agriculture and hydrology. The model takes into account several types of land, family labor, capital to invest and water to irrigate crops. The results help to evaluate the optimal plans under various climate scenarios. lt shows under which conditions various mitigation strategies can improve rainfed production by reducing agricultural risk.