The US Forest Service and other land-management agencies seek better tools for anticipating future expenditures for wildfire suppression. We developed regression models for forecasting US Forest Service suppression spending at 1-, 2-, and 3-year lead times. We compared these models to another readily available forecast model, the 10-year moving average model, and found that the regression models do a better job of forecasting the expenditures for all three time horizons. When evaluated against the historical data, our models were particularly better at forecasting the more recent years (2000-2007) than the less sophisticated models. The regression models also allowed us to generate, using simulation methods, forecast statistics such as the means, medians, and confidence intervals of costs. These additional statistics provide policymakers, wildfire managers, and planners more information than a single forecast value.