Post-fire shifts in vegetation composition will have broad ecological impacts. However, information characterizing post-fire recovery patterns and their drivers are lacking over large spatial extents. In this analysis we used Landsat imagery collected when snow cover (SCS) was present, in combination with growing season (GS) imagery, to distinguish evergreen vegetation from deciduous vegetation. We sought to (1) characterize patterns in the rate of post-fire, dual season Normalized Difference Vegetation Index (NDVI) across the region, (2) relate remotely sensed patterns to field-measured patterns of re-vegetation, and (3) identify seasonally-specific drivers of post-fire rates of NDVI recovery. Rates of post-fire NDVI recovery were calculated for both the GS and SCS for more than 12,500 burned points across the western United States. Points were partitioned into faster and slower rates of NDVI recovery using thresholds derived from field plot data (n=230) and their associated rates of NDVI recovery. We found plots with conifer saplings had significantly higher SCS NDVI recovery rates relative to plots without conifer saplings, while plots with ≥50% grass/forbs/shrubs cover had significantly higher GS NDVI recovery rates relative to plots with <50%. GS rates of NDVI recovery were best predicted by burn severity and anomalies in post-fire maximum temperature. SCS NDVI recovery rates were best explained by aridity and growing degree days. This study is the most extensive effort, to date, to track post-fire forest recovery across the western U.S. Isolating patterns and drivers of evergreen recovery from deciduous recovery will enable improved characterization of forest ecological condition across large spatial scales.