The decision whether to seed with native species following restoration treatments should be based on existing vegetation, species present in or absent from the soil seed bank, past management history, microclimate conditions and soils. We installed three permanent monitoring plots in two areas (total 18.6 ha) at Mt. Trumbull, AZ. Trees were thinned and the sites burned in 1996 and 1997. A 5 ha area was seeded with native shrub, grass and forb species; the remaining 13.6 ha were unseeded. Pretreatment species richness ranged from none to five species per plot. We recorded 13 graminoid and eight shrub species in the seeded area, and four graminoid and four shrub species in the unseeded area. The greatest increase in species richness in both seeded and unseeded plots occurred approximately 1.8 years post-treatment. Perennial native species dominated plant cover by 2.8 years, although annual native forbs dominate the soil seed bank. Perennial grasses are nearly absent from the seed bank. The seeded area had the highest diversity, but it also had twice as many nonnative species (14 versus 7 in the unseeded plots). By August 1999, maximum species richness reached 51 species on the seeded plot. Of these species, 80 percent were native. Although seeding increases diversity, it may also have the long-term tradeoff of introducing new genotypes and species, both native and nonnative.