Wildfires are a growing management concern in western US rangelands, where invasive annual grasses have altered fire regimes and contributed to an increased incidence of catastrophic large wildfires. Fire activity in arid, nonforested ecosystems is thought to be largely controlled by interannual variation in fuel amount, which in turn is controlled by antecedent weather. Thus, long-range forecasting of fire activity in rangelands should be feasible given annual estimates of fuel quantity. Using a 32-yr time series of spatial data, we employed machine learning algorithms to predict the relative probability of large (> 405 ha) wildfire in the Great Basin based on fine-scale annual and 16-d estimates of cover and production of vegetation functional groups, weather, and multitemporal scale drought indices. We evaluated the predictive utility of these models with a leave-1-yr-out cross-validation, building spatial hindcasts of fire probability for each year that we compared against actual footprints of large wildfires. Herbaceous aboveground biomass production, bare ground cover, and long-term drought indices were the most important predictors of burning. Across 32 fire seasons, 88% of the area burned in large wildfires coincided with the upper 3 deciles of predicted fire probabilities. At the scale of the Great Basin, several metrics of fire activity were moderately to strongly correlated with average fire probability, including total area burned in large wildfires, number of large wildfires, and maximum fire size. Our findings show that recent years of exceptional fire activity in the Great Basin were predictable based on antecedent weather-driven growth of fine fuels and reveal a significant increasing trend in fire probability over the past 3 decades driven by widespread changes in fine fuel characteristics.