Immediately after reconstruction of the anterior cruciate ligament, 110 patients were randomly assigned to treatment with high-intensity neuromuscular electrical stimulation (thirty-one patients), high-level volitional exercise (thirty-four patients), low-intensity neuromuscular electrical stimulation (twenty-five patients), or combined high and low-intensity neuromuscular electrical stimulation (twenty patients). All treatment was performed isometrically with the knee in 65 degrees of flexion. All of the patients participated in an intensive program of closed-kinetic-chain exercise. After four weeks of treatment, the strength of the quadriceps femoris muscle and the kinematics of the knee during stance phase were measured. Quadriceps strength averaged 70 per cent or more of the strength on the uninvolved side in the two groups that were treated with high-intensity electrical stimulation (either alone or combined with low-intensity electrical stimulation), 57 per cent in the group that was treated with high-level volitional exercise, and 51 per cent in the group that was treated with low-intensity electrical stimulation. The kinematics of the knee joint were directly and significantly (p < 0.05) correlated with the strength of the quadriceps. There was a clinically and statistically significant (p < 0.05) difference in the recovery of the quadriceps and the gait parameters according to the type of operation that had been performed: the patients who had had reconstruction of the anterior cruciate ligament with use of an autologous patellar-ligament graft did poorly compared with the other patients.