Patterns of vascular plant species richness were investigated in six land use legacy sites (2 formerly plowed, 2 formerly pastured, and 2 permanent woodlot) in Prospect Hill to test predictions about the effects of disturbance, light and soil resources, and forest floor environmental heterogeneity on community composition. The occurrence of vascular taxa was recorded in each of the 60 5 m x 5 m contiguous plots within the 30 m x 50 m permanently gridded study plot in each land use legacy site in June 1996. Identification was made to species in most cases. Woodlots showed higher average species richness at the site level (53) than either pastured (52 species) or plowed (49) sites. However woodlots also show greater spatial variation in richness at the 5 m x 5 m resolution than either the plowed or pastured sites (in that order) as represented by the range, standard deviation, and coefficient of variation. Seasonally-averaged light levels at 50 cm are approximately twice as high in the plowed and pastured sites as in the woodlots, but the woodlots show significantly greater soil organic matter, carbon, nitrogen and water-holding capacity than the post-agricultural sites. These results suggest that soil resources may be more important than light in fostering higher herbaceous stratum richness. However, woodlot richness is also affected by the presence of taxa that are slow to re-colonize heavily disturbed sites (e.g., Epigaea repens) and to distinctive microsites that are less common in the plowed or pastured sites in the Harvard Forest system (e.g., exposed boulders). The greater spatial variation in woodlot richness is strongly influenced by both substrate diversity and by scattered hemlock trees, which substantially depress the herb stratum in localized patches.