Most of the large scale modeling systems used in the analysis of agricultural policies produce deterministic projections. In reality, however, the agricultural sector is subject to a high degree of uncertainty as a result of fluctuations in exogenous factors such as the weather or macroeconomic variation. A stochastic approach can provide additional information to policy makers regarding the implications of this uncertainty, through the use of stochastically generated projections. This paper also shows how deterministic analysis may result in systematic errors in the projection of some variables. As an applied example, the FAPRI model of the US agricultural sector is simulated stochastically to analyse the impact of proposals for the new US farm bill.