We characterized plant regeneration in four old logging roads (700-1000 m long), 12-17 yr after abandonment, in selectively logged forests in lowland Costa Rica. Sets of 4-m2 plots were laid out at 20-m intervals in three distinct microhabitats: road track (topsoil eliminated), road edge (where removed topsoil accumulates on the sides after road construction), and adjacent logged forest. Density of stems ≥ 1 m tall and ≤5 cm DBH (included canopy tree, midstory tree, liana, palm, shrub, and tree fern species) was highest in the road edge plots than either the track or logged forest plots. This "edge effect" is presumably due to buried seed germination of light-demanding trees and shrubs after moderate soil disturbance, less compaction, and higher substrate fertility than in road tracks. Species richness was the lowest, but relative dominance the highest, in the track plots of all roads: 6-9 species comprised alone 50 percent of the Importance Value Index (IVI), in contrast to 11-15 and 16-22 species required to reach 50 percent IVI in edge and forest plots, respectively. We found evidence of soil compaction in tracks of three out of four roads which, in addition to low substrate fertility, and initial lack of on-site plant propagules, could explain slower recovery of stem density and species richness compared to edge and logged forest plots. For stems >5 cm and ≤20 cm DBH, density and basal area in the track plots averaged about one-fourth of edge and logged forest plot values. We estimated recovery of basal area in road tracks to take at least 80 yr to reach the status found in logged forest, and species richness over an even longer period. We suggest that abandoned logging roads serve as long corridors of relatively uniform and long-lasting floristic and structural characteristics that may confer particular ecological roles in selectively logged forests.