This study aims to determine the outcome of double-bundle anterior cruciate ligament (ACL) reconstruction using an allograft in comparison with ACL reconstruction using a double-bundle autograft or a single-bundle allograft. A total of 424 patients who accepted primary ACL reconstructions were divided randomly into three groups: double-bundle technique with autograft (DB-AU group, n = 154), double-bundle technique with allograft (DB-AL group, n = 128), and single-bundle technique with allograft (SB group, n = 142). The KT-1000 arthrometer and pivot-shift tests were performed at 3, 12, and 36 months after surgery, and clinical outcome measurements include the Lysholm score and the IKDC rating scales. Radiological assessments evaluated arthritic changes and tunnel expansion at 36 months postoperatively. The KT-1000 test scores in the DB-AU and DB-AL groups were significantly better than those in the SB group at 12 and 36 months postoperatively (P < 0.05). The pivot-shift tests scores in the DB-AU and DB-AL groups were significantly better than those in the SB group at the 3, 12, and 36 month follow-ups (P < 0.05). Based on the IKDC score and Lysholm score, there were no significant difference between the three groups during follow-up (P > 0.05). At 36 months postoperatively, 42.3 % of patients in the SB group showed a progression in arthritic changes, which was greater than in the DB-AU (29.2 %) and DB-AL (27.3 %) groups (P < 0.05). At 36 months, the rates of tunnel expansion in the DB-AU group and the DB-AL group were lower than in the SB group (P < 0.05). Double-bundle ACL reconstruction can be used to achieve better anterior and rotational stability and has a lower rate of arthritic progression and tunnel expansion than the single-bundle procedure. I.