Revegetation of degraded arid lands often involves supplementing impoverished seed banks and improving the seedbed, yet these approaches frequently fail. To understand these failures, we tracked the fates of seeds for six shrub species that were broadcast across two contrasting surface disturbances common to the Mojave Desert—sites compacted by concentrated vehicle use and trenched sites where topsoil and subsurface soils were mixed. We evaluated seedbed treatments that enhance soil‐seed contact (tackifier) and create surface roughness while reducing soil bulk density (harrowing). We also explored whether seed harvesting by granivores and seedling suppression by non‐native annuals influence the success of broadcast seeding in revegetating degraded shrublands. Ten weeks after treatments, seeds readily moved off of experimental plots in untreated compacted sites, but seed movements were reduced 32% by tackifier and 55% through harrowing. Harrowing promoted seedling emergence in compacted sites, particularly for the early‐colonizing species Encelia farinosa, but tackifier was largely ineffective. The inherent surface roughness of trenched sites retained three times the number of seeds than compacted sites, but soil mixing during trench development likely altered the suitability of the seedbed thus resulting in poor seedling emergence. Non‐native annuals had little influence on seed fates during our study. In contrast, the prevalence of harvester ants increased seed removal on compacted sites, whereas rodent activity influenced removal on trenched sites. Future success of broadcast seeding in arid lands depends on evaluating disturbance characteristics prior to seeding and selecting appropriate species and seasons for application.