Woody plant encroachment has degraded grassland and savanna ecosystems worldwide by decreasing herbaceous production and diversity, and altering these physiognomies toward woodlands. This study evaluated the long-term efficacy of fire and herbicide restoration strategies used in the southern Great Plains to reduce Honey mesquite (Prosopis glandulosa) dominance, restore a grassland/savanna physiognomy, and increase herbaceous production and diversity. Three treatments were evaluated: high-intensity winter fire, aerial spray of clopyralid + triclopyr (C + T), and aerial spray of clopyralid and were compared to untreated mesquite woodland (control). Post-treatment mesquite stand physiognomy was different between fire (low mortality, high basal sprouting), C + T (high mortality, high basal sprouting of surviving plants), and clopyralid (moderate mortality, low basal sprouting of surviving plants) treatments. From 6 to 8 years post-treatment, herbaceous production was increased in C + T and clopyralid treatments but not in the fire treatment. Mesquite regrowth in the fire treatment exerted a competitive influence that limited herbaceous production. Herbaceous functional group diversity was increased in fire and C + T treatments due to a decrease in C3 perennial grass dominance and an increase in C4 perennial grasses and/or C3 forbs. Treatments that maintained mesquite overstory (control and clopyralid) had lower herbaceous diversity due to C3 perennial grass dominance and lower C4 perennial grass cover. The clopyralid treatment demonstrated greatest potential for long-term restoration of southern Great Plains savanna by reducing mesquite canopy cover to historic levels, limiting mesquite basal regrowth and increasing grass production.