This study proposed two futures-based models for forecasting cash prices of corn, soybeans, wheat and cotton over the period 2000-2016. The difference model predicts changes in cash prices as a function of changes in futures prices. The regime model specifies different market regimes and models cash price levels based on observed futures prices in various regimes. The out-of-sample performance of both models was compared to the benchmark of a 5-year moving average over the 2013-2016 sub-period. Our results suggest that the regime model performed best for corn and soybeans. While, neither model beat the benchmark for wheat at the longer forecasted horizons, the difference model performed well at short forecast horizons (up to 5-months ahead). Both models performed better than the benchmark for cotton price forecasts, but they were not significantly different from each other.