For the semiarid Sahelian region, climate variability is one of the most important risks of food insecurity. Field experimentations as well as crop modeling are helpful tools for the monitoring and the understanding of yields at local scale. However, extrapolation of these methods at a regional scale remains a demanding task. Remote sensing observations appear as a good alternative or addition to existing crop monitoring systems. In this study, a new approach based on the combination of vegetation and thermal indices for rainfed cereal yield assessment in the Sahelian region was investigated. Empirical statistical models were developed between MODIS NDVI and LST variables and the crop model SARRA-H simulated aboveground biomass and harvest index in order to assess each component of the yield equation. The resulting model was successfully applied at the Niamey Square Degree (NSD) site scale with yield estimations close to the official agricultural statistics of Niger for a period of 11 years (2000–2011) ($r = 0.82,;text{p-value} < 0.05$). The combined NDVI and LST indices-based model was found to clearly outperform the model based on NDVI alone ($r = 0.59,;text{p-value} < 0.10$). In areas where access to ground measurements is difficult, a simple, robust, and timely satellite-based model combining vegetation and thermal indices from MODIS and calibrated using crop model outputs can be pertinent. In particular, such a model can provide an assessment of the year-to-year yield variability shortly after harvest for regions with agronomic and climate characteristics close to those of the NSD study area.