In a prospective eight-year study comparing surgical and conservative treatment of acute tears of the anterior cruciate ligament, ninety-two patients in whom a tear had been confirmed either by arthroscopy or by arthrotomy were treated and could be followed. The choice of treatment was determined entirely by whether the result of a pivot-shift test was graded as absent, trace, or mild (non-operative treatment) or as moderate or severe (surgical treatment). Surgical treatment consisted of primary repair of the torn ligament and augmentation with a patellar tendon graft. Twenty-two patients were treated non-operatively, and the results were evaluated after twenty-four to eighty-two months (average, forty-eight months). Seventy patients were treated surgically, and fifty-two of them (approximately 70 per cent) returned for follow-up after twenty-four to 100 months (average, forty-eight months); an additional eighteen patients answered a questionnaire. In the non-operative group, about half of the results were graded as excellent or good and half, as fair or a failure. In the surgical group, all but two of the patients had an excellent or a good result; two patients had a fair result. No result was graded as a failure. The results of this study suggest that when the pivot-shift test is not strongly positive, half of the patients will do reasonably well after treatment with a non-operative program of functional rehabilitation. The patients in this study who had a more unstable knee had far better results after a repair and augmentation procedure than have been previously reported after primary repair alone.