The purpose of this study was to clarify differences in tendon graft-to-bone tunnel healing between bone-attached tendon grafts with interference-screw fixation and bone-free tendon grafts with extra-articular suture fixation. In 42 Japanese White rabbits, anterior half replacement of the medial collateral ligament was performed using half of the ipsilateral patellar tendon. At the femoral attachment, the bone-plug-attached graft was fixed with an interference screw (group A). The bone-plug-free graft was fixed by the extra-articular suture fixation technique with sutures tied over a button (group B). Biomechanical and histological evaluations were performed at 2, 4 and 8 weeks postoperatively. In biomechanical evaluation, at 2 or 4 weeks 27 of 28 specimens (96%) were pulled out from the femoral tunnel, while one 4-week specimen and all four 8-week specimens failed at the graft's mid-substance. At 2 weeks, the maximum failure load was 25+/-10 N and 24+/-6 N for group A and group B respectively (mean+/-SD). At 4 weeks, the maximum failure load was 42+/-17 N and 35+/-15 N respectively. There were no significant differences in maximum pullout failure load between the groups at 2 or 4 weeks postoperatively. (P=0.887 at 2 weeks and P=0.339 at 4 weeks using ANOVA measurement). Histologically, the bone-attached grafts showed partial bone-to-bone union at the graft-bone tunnel interface at 4 weeks, and complete bony union at 8 weeks. The bone-free grafts exhibited newly formed Sharpey-like collagen fibers at 4 weeks, and strong connection by mature granulation tissue at 8 weeks. Graft-to-bone tunnel healing of bone-attached graft with screw fixation and bone-free graft with extra-articular suture fixation are comparable in terms of biomechanical evaluation during the early postoperative periods.