A total of 218 patients with unilateral anterior cruciate ligament deficiency were randomly assigned to one of four groups. In group A an anatomical double bundle anterior cruciate ligament reconstruction was performed; group B were treated by a single bundle using an Endobutton for femoral fixation; in group C by a single bundle using RigidFix cross pins for femoral fixation; and in group D by a single bundle using a bioabsorbable TransFix II screw for femoral fixation. For tibial fixation a bioabsorbable Intrafix interference screw was used for all the groups and the graft was fashioned from the semitendinosus and gracilis tendons in all patients. In all, 18 patients were lost to follow-up. The remaining 200 were subjected to a clinical evaluation, with assessment of the anterior drawer, Lachman's and the pivot-shift tests, and KT-1000 arthrometer measurement. They also completed the International Knee Documentation Committee, Lysholm knee and Tegner activity scores. At a mean of 29 months (25 to 38) follow-up there were no significant differences concerning time between injury and range of movement and Lysholm knee scores among the four groups. However, the double bundle method showed significantly better results for the pivot-shift test (p = 0.002). The KT 1000 measurements showed a mean difference between the reconstructed knee and the patients' normal knee of 1.4 mm in the double bundle group and 2.4 mm in the single bundle group; which was statistically significant. The Lachman and anterior drawer tests also showed superior results for the double bundle method. The International Knee Documentation Committee scale showed no significant difference among the groups (p < 0.001). On clinical evaluation the double bundle group showed less laxity than the single bundle groups. However, regardless of the technique, all knees were improved by anterior cruciate ligament reconstruction compared with their pre-operative status.