Eastern Virginia grain producers face production and marketing decisions under conditions of considerable yield and price variability. Traditionally MOTAD and other risk programming models have relied on the variability of historical returns series as a proxy for the risk an individual producer experiences. The mean forecast deviation method provides an alternative to the standard method of calculating deviations for MOT AD models. This method was applied to an eastern Virginia cash grain farm to provide a farm plan based on expected prices during the first week of February. The acreages of com, soybeans, and small grains were specified based on soil type, government program participation, sales at harvest, storage, pre-harvest hedging, and post-harvest hedging. The expected farm plans for the 1987 crop year resulted in participation in the government commodity programs at all levels of risk aversion and for each crop and soil type. Elimination of government commodity programs from the model resulted in two major findings: wheat and barley doubled cropped with soybeans preformed well on both soil types and at all levels of risk aversion, and given 1987 expected prices, idling lower quality land was a profitable decision at all levels of risk aversion. / Master of Science