A model of local elevator basis levels was developed to quantify the response of corn, soybean and hard red spring wheat basis to a set of predefined predictors. Basis data from 2013 – 2016 for 12 grain elevators in Eastern North Dakota were collected. A maximum likelihood mixed effect model was used to test the significance of alternative predictor variables and further divided the residuals into temporal and spatial components. The results indicate that the base model was able to explain 57 to 87 percent of the local basis variability for the selected crops at the individual elevators. In addition, the findings suggest that the temporal portion of the remaining variability is greater than the spatial variability for corn and soybean, but similar for hard red spring wheat.