Encroachment of woody plants into grasslands, and subsequent brush management, are among the most prominent changes to occur in arid and semiarid systems over the past century. Despite the resulting widespread changes in landcover, substantial uncertainty about the biogeochemical impacts of woody proliferation and brush management exists. We explored the role of shrub encroachment and brush management on leaf litter decomposition in a semidesert grassland where velvet mesquite (Prosopis velutina) abundance has increased over the past 100 years. This change in physiognomy may affect decomposition directly, through altered litter quality or quantity, and indirectly through altered canopy structure. To assess the direct and indirect impacts of shrubs on decomposition, we quantified changes in mass, nitrogen, and carbon in litterbags deployed under mesquite canopies and in intercanopy zones. Litterbags contained foliage from mesquite and Lehmann lovegrass (Eragrostis lehmanniana), a widespread, nonnative grass in southern Arizona. To explore short- and long-term influences of brush management on the initial stages of decomposition, litterbags were deployed at sites where mesquite canopies were removed three weeks, 45 years, or 70 years prior to study initiation. Mesquite litter decomposed more rapidly than lovegrass, but negative indirect influences of mesquite canopies counteracted positive direct effects. Decomposition was positively correlated with soil infiltration into litterbags, which varied with microsite placement, and was lowest under canopies. Low under-canopy decomposition was ostensibly due to decreased soil movement associated with high under-canopy herbaceous biomass. Decomposition rates where canopies were removed three weeks prior to study initiation were comparable to those beneath intact canopies, suggesting that decomposition was driven by mesquite legacy effects on herbaceous cover-soil movement linkages. Decomposition rates where shrubs were removed 45 and 70 years prior to study initiation were comparable to intercanopy rates, suggesting that legacy effects persist less than 45 years. Accurate decomposition modeling has proved challenging in arid and semiarid systems but is critical to understanding biogeochemical responses to woody encroachment and brush management. Predicting brush-management effects on decomposition will require information on shrub-grass interactions and herbaceous biomass influences on soil movement at decadal timescales. Inclusion of microsite factors controlling soil accumulation on litter would improve the predictive capability of decomposition models.