The genetic diversity and structure of Quercus rubra L. (Red Oak) in five secondary populations regenerated on old pastures and five paired old-growth populations were compared to determine whether secondary populations may have lost genetic diversity present in the presettlement forest. Old-growth samples had significantly greater nuclear allelic richness than secondary stands at the four microsatellite, simple sequence repeat (SSR), markers tested, but there was no significant difference at three chloroplast SSRs. There was however tenfold greater differentiation (high F_ST) among secondary forests than among old-growth forests for the chloroplast loci, suggesting greater population isolation at the time of their founding. The young age class in secondary stands had significantly greater chloroplast allelic richness than the old age class, suggesting some recovery as the forest cover increased. Forest cover and patch size at peak deforestation was not correlated with allelic richness in secondary stands, but the power of this test was low. The current size of old-growth stands is, however, positively correlated with levels of genetic diversity. Results suggest that deforestation for agriculture may reduce genetic diversity of forest trees, particularly in the first generation after deforestation, but that some recovery of this diversity is likely in these patchy landscapes. Wind dispersed pollen flow may increase in these areas, but animal dispersed seed flow may be reduced.