Remotely sensed data have been used in the past to predict crop yields. This research attempts to incorporate remotely sensed data into a net farm income projection model. Using in-sample regressions, satellite imagery appears to increase prediction accuracy in the time periods prior to USDA's first crop production estimate for wheat and corn. Remotely sensed data improved model performance more in the western regions of the state than in the eastern regions. However, in a jackknife out-of-sample framework, the satellite imagery appeared to statistically improve only 8 of the 81 models (9 crop reporting districts by 9 forecasting horizons) estimated. Moreover, 41 of the 81 models were statistically better without the satellite imagery data. This indicates that perhaps the functional form of net farm income has not been well-specified since additional information should generally not cause a model to deteriorate.