Crop county estimates, produced by the USDA's National Agricultural Statistics Service (NASS), are used by the private sector; colleges and universities; and local, state and national governments to monitor shifts in agricultural production. These estimates are based on a non-probability sample of farming operations in a state, so, as Stasny, Goel, et. al. (1995) point out, they cannot be generated using traditional estimation methods based on known selection probabilities. Statisticians at The Ohio State University, under a cooperative agreement with NASS, have developed a county yield estimation algorithm based on the assumption that neighboring counties have similar yields. The algorithm is a mixed-effects model which uses post-stratification on total land operated and allows for differences in yield by county’ and farm size. This report presents the results of a simulation study designed to assess the effectiveness and reliability of the algorithm. The results indicate that the algorithm produces improved yield estimates when compared to standard ratio estimates. This paper briefly discusses the spatial methodology used in the model, explores potential problems with the algorithm, examines the results of preliminary testing and suggests several steps that should be taken in preparation for operational implementation.