The FSim wildfire simulation model is widely used to generate estimates of burn probability (BP). However, few studies have compared BP models to subsequent wildfires to assess their suitability for estimating near-future wildfire risk. Here, we compared the U.S.D.A. Forest Service’s publicly available BP map for the conterminous U.S. (Short et al., 2020; Dillon et al., 2023) to observed wildfire perimeters. Our main focus was to evaluate the BP map version based on 2014 LANDFIRE fuels data and calibrated to historical wildfires from 1992-2015, allowing us to compare BP to observed wildfires from 2016-2022. We also compared evaluations using a newer version of the BP map based on 2020 LANDFIRE fuels and 1992-2020 historical wildfires, and additionally performed evaluations for the western U.S. based on differing wildfire size classes. This dataset includes CSV tables summarizing burned area proportions by BP classes for individual pyromes, or regions with similar wildfire regimes, as well as processing code used to summarize burned area by BP classes.