Intra‐articular soft tissues, such as the anterior cruciate ligament (ACL), fail to heal in contrast to the extra‐articular medial collateral ligament (MCL), which undergoes classic healing. The goal of this study was to validate a model for failure of intra‐articular healing that could be used in the future to test new repair strategies. We conducted a two‐part experiment, the first part ex vivo, and the second in vivo. Our initial ex vivo experiments were used to determine the optimal width of the central defect in the canine ACL that would produce reproducible structural properties at time zero. The second experimental series used this optimal scalpel blade width to create a central defect in the canine ACL followed by measurement of structural properties in the ACL after either a 3‐ or 6‐week in vivo healing period. A 3.5‐mm beaver blade resulted in a maximum tolerated load of 56.8 ± 4.7% (mean ± SEM) of control at time zero. After the 3‐ and 6‐week in vivo healing periods, the maximum load was 74.6 ± 5.3 at 3 weeks and 64.9 ± 3.8% at 6 weeks compared to control. Thus, biomechanical parameters tested at 6 weeks after creation of a defect showed no significant gains from defects tested immediately after the creation of injury. The centrally placed ACL defect in this canine model demonstrates failure to mechanically heal, which should prove suitable for future in vivo evaluation of the biomechanical and histological response to tissue engineering repair strategies for intra‐articular soft tissues. © 2006 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 24:401–406, 2006