We have studied the results of reconstruction by freeze-dried patellar allografts or patellar autografts in ACL-deficient patients prospectively. From January 1995 to December 1995, we performed ACL reconstruction using an arthroscopic-assisted technique with patellar autografts in 21 patients and patellar allografts in 13 patients. Minimum followup time was 1 year(average 26 months). All patients were evaluated by using KT-2000 arthrometer and MRI as well as by physical examination. Final results were rated as satisfactory or unsatisfactory by using a modified Feagin knee scoring scale. Good or excellent were considered to have satisfactory results and fair or poor were considered to have unsatisfactory results. As measured by the KT-2000, 19 cases$(90.5\%)$ had a 5-mm or Jess side-to-side differential, a satisfactory results in autograft group, 2 cases of unsatisfactory results had joint instability. In allograft group, 10 cases$(76.9\%)$ had a 5-mm or less side-to-side differential, a satisfactory results, 3 cases of unsatisfactory results had joint instability including postoperative infection(1 case). In conclusion, the results of ACL reconstruction with autografts were better than those with allografts. The problem of allograft reconstruction were rehydration, aseptic control and improper mechanical tensioning. So, we thought that success of allograft reconstruction was depended on careful implant preparation including pretensioning technique.