Urban forests represent patches of biodiversity within otherwise degraded landscapes, yet these forests are threatened by invasion by exotic plant species. We investigated the response of a forest understory to removal of four common exotic species: Elaeagnus umbellata Thunb., Lonicera japonica Thunb., Ligustrum sinense, Laur., and Microstegium vimineum (Trin.) A. Camus in a forest within the city of Raleigh, NC, USA. In the summer of 2001, we initiated a removal experiment with three treatments. In the “repeated removal” treatment, all understory vegetation was initially removed by clipping and new exotic seedlings were repeatedly removed every 2 weeks throughout the study period. The “initial removal” treatment involved a one-time understory vegetation removal with no further weeding. Control plots had no intervention throughout the study period. We conducted vegetation surveys of the plots prior to treatment initiation and in April and August of 2002 and 2003. With a non-metric multidimensional scaling (NMS) ordination, we were able to discern differences in species composition between the repeated removal treatment and the other two treatments. However, using repeated measures ANOVA, we found no significant differences in native species richness, cover, and abundance among treatments during most sampling periods. We also used a seedbank study to determine that while some early successional species were present, no native shrubs and few native trees emerged from the seedbank. These results suggest that (1) repeated removal is required to decrease the importance of exotic species, especially if the site is in close proximity to a source of exotic propagules; and (2) subsequent to exotic removal, native species may not recover sufficiently without supplemental plantings. Therefore, restoration plans for urban forests should incorporate both long-term monitoring and native plant re-introduction to achieve a diverse native community.