Woody plant encroachment is a pervasive global driver impacting grassland structure and functioning, and we know very little about efficacies of management techniques for restoring woody plant-encroached tropical grasslands. In the Banni grasslands of India, which are extensively encroached by an invasive woody legume, Prosopis juliflora, we test the efficacies of two Prosopis management methods-mechanical removal and lopping-to restore native grasslands. Across eight sites in the Banni, using a randomized block design, we monitored herbaceous plant recovery in the two Prosopis management treatments, and an unmanipulated control for 4 years. We also sampled four reference grasslands. We found native herbaceous plant richness and cover to be threefold (mean = 6 species) and sixfold (mean = 18% cover) higher, respectively, in the mechanical-removal treatment compared to the control (mean richness = 2 species, cover = 3%), but found no significant differences between the lopping treatment (mean richness = 5 species, cover = 11%) and the control. While mechanical-removal increased plant diversity, neither management treatment changed the species composition, and both of these treatments were still missing perennial grasses that dominate the reference grasslands. Our data highlight the efficacy of mechanical removal over lopping in restoring native grasslands. However, mechanically removing Prosopis is costly, especially at larger scales. Apart from its detrimental impact on grasslands, Prosopis also has benefits varying with stakeholder groups, and its removal involves complex trade-offs. Until we fully understand the trade-offs involved, it may therefore be better to manage this social-ecological landscape as a mosaic of restored grassland and Prosopis-invaded patches.