A geographic information system model designed to identify regions at risk for West Nile virus (WNV) transmission was calibrated and tested with data collected in Santa Clara County, California. American Crows that died from WNV infection in 2005 provided spatial and temporal ground truth. When the model was run with parameters based on Culex tarsalis infected with the NY99 genotype of the virus, it underestimated WNV occurrence in Santa Clara Co. The parameters were calibrated to fit the field data by reducing the number of degree-days necessary to reach the mosquito's extrinsic incubation period from 109 to 76. The calibration raised model efficiency from 61% to 92% accuracy, and the model performed well the following year in Santa Clara Co.