Land managers frequently apply vegetation removal and seeding treatments to restore ecosystem function following woody plant encroachment, invasive species spread, and wildfire. However, the long‐term outcome of these treatments is unclear due to a lack of widespread monitoring. We quantified how vegetation removal (via wildfire or management) with or without seeding and environmental conditions related to plant community composition change over time in 491 sites across the intermountain western United States. Most community metrics took over 10 years to reach baseline conditions posttreatment, with the slowest recovery observed for native perennial cover. Total cover was initially higher in sites with seeding after vegetation removal than sites with vegetation removal alone, but increased faster in sites with vegetation removal only. Seeding after vegetation removal was associated with rapidly increasing non‐native perennial cover and decreasing non‐native annual cover. Native perennial cover increased in vegetation removal sites irrespective of seeding and was suppressed by increasing non‐native perennial cover. Seeding was associated with higher non‐native richness across the monitoring period as well as initially higher, then declining, total and native species richness. Several cover and richness recovery metrics were positively associated with mean annual precipitation and negatively associated with mean annual temperature, whereas relationships with weather extremes depended on the lag time and season. Our results suggest that key plant groups, such as native perennials and non‐native annuals, respond to restoration treatments at divergent timescales and with different sensitivities to climate and weather variation.