The purpose of this study was to review our results with axillofemoral by-passes performed for aorto-iliac occlusive disease. Fifty patients receiving 51 axillofemoral by-passes from January 1989 to December 1994 were retrospectively reviewed. The 30-day post-operative mortality was 4%. Seven patients (14%) presented graft-related local complications and all but one required reoperation. Five patients were lost to follow-up, the mean length of which was 36 months (16-74 months). Forty-nine per cent of the patients died during the follow-up period. At 36 months, the primary patency rate was 51%, the secondary patency rate was 69%, and limb valvage rate was 87%. A statistical difference was seen in the secondary patency rate between axillobifemoral by-pass (87%) and axillo-unifemoral by-pass (56%) at 36 months (P < 0.01), but no difference was seen in the limb salvage rate at 36 months between the two configurations of the by-pass (94% vs 81%) (P = NS). Twenty patients (40%) operated upon for acute ischemia had a significantly higher post-operative mortality rate (10% vs 0), a significantly higher amputation rate (20% vs 6.6%) and a significantly lower patency rate of by-pass (26% vs 63%) (P < 0.01), than the 30 patients (60%) operated on for claudication, rest pain or trophic ulcers. Our findings indicate that the results of axillofemoral by-pass are significantly influenced by the selection of patients for operation, namely the clinical status of ischaemic symptoms, and that since the overall results of axillofemoral by-pass are inferior to those of aortofemoral by-pass, this treatment should be restricted to patients at high risk of aortic clamping.