A significant increase in demand for fuel ethanol in California should be expected if all gasoline sold in the state were to be blended with 10% ethanol, as envisaged in the State Alternative Fuels Plan. This paper assesses the potential of California agriculture to supply biofuel feedstock in the form of switchgrass. We construct a fully calibrated, multi-region, multi-input and multi-output model of agricultural supply for California’s Central Valley based on the principles of Positive Mathematical Programming. We exploit the biogeochemical model DAYCENT to estimate production functions for switchgrass in each agricultural region. We then predict the extent and location of potential switchgrass production in the Central Valley. Our results suggest that adoption rates differ widely among regions, meaning that the location of processing plants may be an important issue. They also suggest that switchgrass adoption is not likely to displace specialty crops by much. From a purely methodological standpoint, this study illustrates the complementarity of agronomic and economic information for the calibration of economic optimization models meant to capture farmer behavior at the regional scale.