Wildfires in the west respond strongly to weather and climatic factors, but it is not clear how models of wildfire ignitions and wildfire area differ or what factors significantly affect them. We present panel cross-sectional models of wildfire ignitions and area by ignition type, including arson, accidental, and lightning fires. These models relate fire incidence to weather, demographic variables, ecological variables, historical wildfire in nearby locations, and wildland management activities. Given the autoregressive nature of these models, managers and policy makers can use them to forecast wildfire risks several months in advance. They also may be used to evaluate the efficacy of alternative fuel treatments to reduce such risks. An example of how the models can be used to forecast the severity of the 2004 season will also be shown.