Beyond socio-economic conditions, the main obstacle to agriculture intensification in Sahelian area is the inter-annual and intra-seasonal rainfall variability because it is synonymous of risk and uncertainty (Hansen, 2005). Their impacts may cause food crisis and maintain dependency on imported cereals. This situation calls for decision support tools such as crop models for yields forecasting allowing preventing eventual crisis (Samba, 1998). Various Senegalese institutions (CERAAS, ISRA, DAPS), collaborating with CIRAD and AGRHYMET, have explored the possibility of using the SarraH©CIRAD crop model (Dingkhun et al., 2003) for yield forecasting. The objectives of this study are to characterize development features of 18 farmers' pearl millet "varieties" under optimal water and nutrient conditions aiming at model calibration, and characterize on-farm yields in Diourbel District in order to compare the model output to on-farm fields' data.