Controversy continues as to whether percutaneous transluminal angioplasty (PTA) or surgical revision is the ideal modality for the treatment of failing grafts. This prompted a histopathologic analysis of failing human vein graft segments subjected to ex vivo balloon dilatation to define variables responsible for the discrepant results. Fifteen vein graft lesions from 14 patients with failing infrainguinal bypasses were recovered after surgical excision. Each graft lesion was focal and uniform in length (2.1 ± 0.3 cm). Rings sectioned from adjacent regions of each vein graft lesion before and after balloon inflation were processed for histologic study, photomicrography, and image analysis. Angioplasty balloon size was selected on the basis of preoperative arteriograms. Graft lesions were divided into three groups based on lesion thickness and the degree of fibrosis and cellularity seen on sections stained with Masson's trichrome. The luminal area before angioplasty was not significantly different for the three groups (p >0.2). Vein grafts with thick intimas (group 1) had significantly less luminal dilatation after angioplasty as compared with less thick intimal lesions (groups 2 and 3; p< 0.001). Those lesions with varying degrees of cellularity (groups 2 and 3) showed no significant differences in luminal diameter after angioplasty. However, the cellular lesions in group 2 consistently formed multiple intimal flaps that could produce PTA failures even with good luminal restoration. The varying histology of vein graft lesions and associated differences in intimal thickness and cellularity may be responsible for the inconsistent results following PTA. Estimates of wall thickness before angioplasty, particularly in the intimal area, may be helpful in evaluating which lesions might benefit most from PTA.