This study was carried out to determine whether we could develop a model to identify predictive factors for success of femoral-popliteal (FP) bypass grafts. In a retrospective review of 199 operations, 24 factors influencing outcome were selected by stepwise logistic regression analysis, a sophisticated, multifactorial computer program. The top five indicators (excluding intraoperative technical problems) were runoff status, previous ipsilateral FP bypass, preoperative prediction of potential amputation level, concurrent proximal vascular reconstruction, and site of the distal anastomosis. We chose to validate the predictive model developed before applying it clinically. Data from 67 subsequent cases were presented to the computer without the known outcome, and the probability of 30-day patency was calculated. The model predicted 11 failures; there were actually eight thromboses in the 67 grafts. However, only three of the failures were predicted correctly, and eight cases of computer-anticipated thromboses were patent at 30 days. The high false positive rate makes the clinical application of the predictive model inappropriate. The success of FP bypass grafts appears to be related to factors that cannot be assessed preoperatively, such as technical problems during surgery. Even those which seem to have a poor runoff and other high-risk factors may succeed; thus an aggressive approach is justified in lower extremity reconstructions.