We conducted experiments to examine microsite and rodent influence on mesquite (Prosopis glandulosa) germination and survival in Chihuahuan Desert black-grama (Bouteloua eriopoda) grassland. Simulated caches with mesquite seeds were placed in undisturbed grassland, burned grassland patches with reduced grass cover, and on the periphery of kangaroo rat mounds. Rodent access to cached seeds was controlled by covering half of the caches with mesh domes. Kangaroo rats destroyed dung pats containing scarified mesquite seeds and excavated all seeds in simulated caches during a year in which no grasses or forbs set seed. Germination was the highest in unburned grassland, but over-winter survival was only observed across experiments in burned grassland microsites. A large proportion of seed caches were excavated (1.1–5.5% of the rodent-excluded caches and between 15.6% and 21.1% of the control caches). There were seasonal differences in numbers of emergent seedlings, with a peak in September 2002 following late summer rains. There were no significant differences in numbers of germinants in burned and unburned grassland microsites, but significantly fewer mesquite seeds germinated on banner-tail kangaroo rat mounds. Caches excavated by rodents yielded 46.6% of germinants. This study suggests that scarification of mesquite seeds is not necessary for germination and heteromyid rodents may be responsible for increased mesquite density in grasslands with some mature mesquite plants. The data also suggest that competition between mesquite seedlings and grasses in black-grama grassland is weak because of the physical characteristics of this system.