Farmland values play a major role in the health of the U.S. agricultural economy. In order to better understand the forces affecting land prices, a simultaneous demand and supply model of the agricultural real estate market is constructed;Price and quantity are jointly dependent variables. Independent variables include anticipations about future land prices, or simply expected capital gains. Three different formulations are used to represent these expectations; naive, adaptive, and rational expectations;Under naive expectations, anticipated land values are simply the last observed value. Under adaptive expectations, land value expectations are a weighted average of past levels. The rational expectations hypothesis (REH) asserts that all relevant information is used when forecasts are made. Thus, expected land values are derived from the anticipated values of the independent variables;Two-stage least squares estimation of the model without capital gains provided good results. Correct hypothesized signs and high significance levels for current price (or quantity) in each equation suggest simultaneity is in fact appropriate for the farmland market. Demand was found to be responsive to farmers' debt/equity ratio and off-farm income, average farm size, the Federal Land Bank interest rate, the percentage of rented land, and population. The income return to land was insignificant. Supply was influenced by farmers' income from the farm and debt/equity ratio, the number of farm foreclosures, and yield on preferred stocks, and positively related to the income return to farmland. Ex post simulation of this model provided good results;The addition of capital gains expectations provided the empirical test of the REH. Naive expectations appeared to be the best specification for expected capital gains. Price simulation, the main purpose of the model, was best in this case. Adaptive expectations entered with less disruption, but also less significance. Little if any support was garnered for the validity of the REH as applied to the farm real estate market. The variables showed little statistical significance, often were of incorrect expected sign, and severely altered the other coefficients. An ex ante forecast was run and impact multipliers calculated for the naive expectations model.