A stochastic dynamic programming model is developed to determine optimal replacement intervals and depreciation schedules for a combine on a cash grain farm in north central Montana, where the optimal decision is based on the stochastic nature of winter wheat prices. Empirical results indicate that the decision varies widely depending on the states describing the conditions facing the farm firm. Under normal profitable conditions and ERTA81 tax legislation, suggested replacement is after five years of service, the new asset being depreciated under the accelerated cost recovery system and the investment credit option. Changes to the tax law would tend to smooth out and increase this replacement interval.