Land-use history, recent management, and landscape position influence vegetation at the Rockefeller Experimental Tract (RET), a 40-year-old restoration experiment in northeast Kansas. RET is representative of the prairie-forest ecotone, containing native tallgrass prairie and oak-hickory forest, but unique in having tracts of replanted prairie, seeded in 1957, that have undergone long-term restoration treatments: burned, grazed, mowed, or untreated. A land-use history database for RET was compiled using a geographic information system to integrate historic and contemporary sources of information. Restoration management on the reseeded prairie has had a profound effect on forest development: mowing or burning precluded forest establishment (<3% forest cover), whereas portions of untreated or grazed areas became heavily forested (>97% forest cover). Forest colonization depends upon biotic and edaphic conditions at the time restoration was initiated: for areas replanted to prairie and managed by grazing, forestation was 6% on land in cultivation prior to replanting, 20% on former pastureland, and 98% on land deforested just before replanting. Patterns of forest colonization were also significantly associated with three landscape positions: near existing forest, along water courses, and along ridge tops. Additionally, land-use history analyses showed that the presence of various prairie and forest species resulted from persistence and not from colonization following restoration. Because of the lasting imprint of historic land use on the landscape, our results indicate that it is essential that restoration studies be evaluated within a site-specific historical context.