Free TRAM flap breast reconstruction was performed in 23 patients from july 1993 through november 1995 at the Saint-Louis Hospital in Paris. The surgical team was composed of eight different surgeons. In all cases a delayed breast reconstruction procedure was performed. All patients in this series had previously received radiation therapy and 82.6% patients presented with excess body weight. Decision to perform a free flap procedure was confirmed peroperatively under two conditions. Adequate caliber of the donor and recipient vessels was required, allowing the anastomoses to be performed without magnification in most cases. Preservation of the thoraco-dorsal vessels was the rule so as to allow later use of a latissimus dorsi flap if necessary. Thus in 8 of the 31 cases in which a free flap was initially indicated a pedicled flap was actually performed so as to satisfy the above conditions. All procedures were performed by two surgical teams working simultaneously. Flap harvest met local tissue requirements in all cases. A lateral strip of rectus abdominis muscle and fascia was preserved when this appeared feasible. The abdominal wall was reinforced by prosthetic means in 82.6% of cases. The average operative time was 9 hours. Results were considered satisfactory or very satisfactory in most cases. The complication rate was 39.1%. Among the complications noted were 1 case of partial flap necrosis, 1 case of fat necrosis, 1 abdominal hernia, 1 abdominal bulge (both abdominal complications occurred in patients in whom no prosthetic material was used for abdominal repair). This complication rate also includes revision of the microsurgical anastomoses in 2 cases; in both cases the flap survived completely. This study tends to suggest that the free TRAM flap for breast reconstruction is a reliable technique. It is the authors' belief that it should replace the bipedicled TRAM flap since it combines ample flap vascularization with minimal rectus harvest.