We investigated the long-term outcome of 100 patients 15 years after having been randomly allocated to primary repair (augmented or non-augmented) or non-surgical treatment of an anterior cruciate ligament (ACL) rupture. The subjective outcome was similar between the groups, with no difference regarding activity level and knee-injury and osteoarthritis outcome score but with a slightly lower Lysholm score for the non-surgically treated group. This difference was attributed to more instability symptoms. The radiological osteoarthritis (OA) frequency did not differ between surgically or non-surgically treated patients, but if a meniscectomy was performed, two-thirds of the patients showed OA changes regardless of initial treatment of the ACL. There were significantly more meniscus injuries in patients initially treated non-surgically. One-third of the patients in the non-surgically treated group underwent secondary ACL reconstruction due to instability problems. In this study, ACL repair itself could not reduce the risk of OA nor increase the subjective outcome scores. However, one-third of the non-surgical treated patients were later ACL reconstructed due to instability. The status of the menisci was found to be the most important predictor of developing OA. Early ACL repair and also ACL reconstruction can reduce the risk of secondary meniscus tears. Indirectly this supports the hypothesis that early stabilization of the knee after ACL injury is advantageous for the long-term outcome.