Recently, we established a new mouse model of vein graft arteriosclerosis by grafting vena cava to carotid arteries. In many respects, the morphological features of this murine vascular graft model resemble those of human venous bypass graft disease. Using this model, we studied the effects of local gene transfer of tissue inhibitor of metalloproteinase-2 (TIMP-2) on vein graft remodeling. Mouse isogeneic vessels of the vena caval veins were grafted end to end into carotid arteries, then enveloped with the replication-defective recombinant adenoviruses overexpressing human TIMP-2 (RAdTIMP-2) or β-galactosidase (RAdLacZ) at 1×10¹⁰ plaque-forming units/mL in a total volume of 50 μL, and incubated at room temperature for 20 minutes. In the untreated group, vessel wall thickening was observed as early as 1 week after surgery and progressed to 4- to 10-fold the original thickness in grafted veins at 4 and 8 weeks, respectively. RAdLacZ vector treatment significantly enhanced neointimal lesions at 8 weeks, which was completely blocked by RAdTIMP-2 gene overexpression. Interestingly, RAdTIMP-2 gene transfer resulted in a reduction in vessel diameter of grafted veins compared with ungrafted veins (819±96 versus 624±67 μm, respectively; P<0.05). Maximal β-galactosidase activity was found at 2 weeks and was detectable until 4 weeks after gene transfer. Double immunofluorescence studies demonstrated that cells overexpressing TIMP-2 were mostly localized in the adventitia and were MAC-1-positive monocytes/macrophages but not smooth muscle cells. Furthermore, the activity of matrix metalloproteinases was markedly decreased in the vessel walls treated with RAdTIMP-2 compared with that in the untreated control group and the RAdLacZ-treated group. Thus, this mouse model has been proven to be useful in gene transfer studies. Our findings demonstrate that local TIMP-2 gene transfer significantly reduces vein graft diameter, ie, remodeling to an artery-like vessel via inhibition of matrix metalloproteinase activity.