This study examines the statistical association of wildfire risk with climatic conditions and non-climate variables in 48 continental US states. Because the response variable “wildfire risk” is a fractional variable bounded between zero and one, we use a non-linear panel data model to recognize the bounded nature of the response variable. We estimate the non-linear panel data model (fractional probit) using the Generalized Estimating Equation (GEE) approach to ensure that the parameter estimation is efficient. The statistical model, coupled with the future climates projected by Global Climate Models (GCMs), is then employed to assess the impact of global climate change on wildfire risk. Our regression results show that wildfire risk is positively related to spring, summer, and winter temperatures and human population density whereas it is negatively associated with precipitation. The simulation results based on GCMs and the regression model indicate that climate change will intensify wildfire risk throughout the entire US, especially in the South Central region, posing an increasing wildfire threat and thus calling for more effective wildfire management strategies.