The ability to accurately forecast basis is crucial to risk management strategies employed by many agribusiness firms. Previous research has examined how to effectively use basis forecasts and what factors affect basis, but literature focusing on forecasting basis is sparse. This research evaluates the impact of adopting a time-to-expiration approach, as compared to the more common calendar approach, when forecasting feeder cattle, live cattle, and hog basis. Furthermore, the optimal number of past year's basis levels to include in making basis predictions is evaluated in an out-of-sample framework. Absolute basis forecasts errors are generated for all three commodities and evaluated to determine the signifcance of the two issues mentioned above. Results indicate that basis forecasters should consider using three-year historical averages for feeder cattle and four-year historical averages for live cattle and lean hogs when making basis forecasts. Furthermore, the use of a time-to-expiration method of calculating historical average basis results in very little improvement in basis prediction accuracy compared to the calendar approach.