Recanalization of a totally occluded saphenous vein graft (SVG) using commercially available urokinase from human kidney cells has been shown to be effective, but the duration of infusion and complications such as allergic reactions, bleeding events, and non-Q-wave myocardial infarction have limited its acceptance. Recently, genetic engineering has allowed the synthesis of recombinant urokinase (r-UK). Patients with an occluded SVG from 37 centers were randomized to receive a 6-hour infusion of either low-dose (125,000 IU/hour) or high-dose (350,000 IU/hour) r-UK followed by up to a maximum of 18 hours of r-UK (125,000 IU/hour) via a subselective catheter directly into the occluded vein graft. The primary study end point was final preintervention achievement of Thrombolysis In Myocardial Infarction (TIMI) flow > or = 2 using core angiographic analysis. One hundred seven patients were randomized and 98 received the study drug (low dose 52 patients, high dose 46 patients). TIMI flow > or = 2 after completion of the study drug was higher in the high-dose group (51% vs 24%, p = 0.019). This difference narrowed, but a trend was still evident on the final angiogram after adjunctive mechanical intervention (72% vs 58%, p = 0.254). Bleeding complications were frequent; severe or life-threatening bleeding occurred in 12% of patients on the low dose and 11% of patients on the high dose (p = NS), including 2 intracerebral bleeds, both of which were fatal with 1 in each group. Thus, in patients with an occluded SVG, a randomized trial of direct low-dose versus high-dose r-UK infusion demonstrated increased recanalization rates (TIMI flow > or = 2) in the high-dose arm. Percutaneous revascularization of SVG with r-UK can be accomplished with acceptable success rates, but complications are frequent.