Abstract Aim The assembly of a forest flora on sites abandoned from agriculture is potentially limited by the physical and chemical suitability of soils. This paper describes the post-agricultural development of soil in deciduous forest sites to gauge its importance in limiting a second-growth forest flora. Location Northern Delaware and south-eastern Pennsylvania, USA. Methods A chronosequence of 99 second-growth forest sites was assembled in the Piedmont region. Stand age ranged from 0–20 to > 100 years since canopy closure, as determined from historical aerial photographs. Seventy-four sites are likely to have experienced cultivation before regeneration of forest. The presence of large rocks or pit-and-mound microtopography suggests that the remaining 25 sites were not cultivated, although they were probably used as pasture and/or woodlot. Soil organic content, moisture, pH, and concentrations of extractable potassium, phosphorous, calcium, magnesium and sulphur were measured in samples from each site. Coarse woody debris, size of the largest tree, canopy openness and several microtopographic features were also recorded. Results Most soil variables fell into two separate but internally correlated groups: the first group including organic matter, moisture and sulphur; the second including pH, calcium, potassium and magnesium. In formerly cultivated sites, organic matter, moisture and sulphur increased gradually through the chronosequence, approaching values recorded in uncultivated sites at 81–100 years. Calcium, potassium, magnesium and pH declined sharply between the 21–40- and 41–60-year classes – probably an indirect effect related to geographic location and selective abandonment. Phosphorous also declined with time, but the decline was not as clearly defined. Main conclusions The progress of soil recovery observed here is faster than rates implied in other studies, possibly because these sites experienced a period as pasture before reversion to forest. Sites appear to have returned to undisturbed-forest levels as fast as or faster than the predicted accumulation of a forest flora, suggesting that assembly of second-growth forest communities is not limited by the rate of soil development in this region.