The knee is the second-most common location for osteonecrosis, although it is affected much less often than the hip. Core decompression by precise drilling into ischemic lesions of the femoral condyle while remaining extra-articular is a challenge, particularly in obese patients. For cell therapy, exact localization of the injection point is important to avoid intra-articular injection.The precision of drilling with computer-based navigation was compared to that of conventional fluoroscopy-based drilling. A prospective, randomized study was conducted using both surgical trainees without experience and expert surgeons. First, participants performed the surgical task (core decompression) on a cadaver knee using fluoroscopic guidance or computer-based navigation. Performance was determined by the radiographic analysis of trocar placement. Next, 12 consecutive patients with bilateral symptomatic secondary (corticosteroids) osteonecrosis without collapse were included in a clinical prospective, randomized, controlled study. The 24 knees were treated using conventional fluoroscopy with expert surgeons on one side and computer-based navigation with surgical trainees on the contralateral side. Bone marrow aspirated from the two iliac crests was mixed before concentration. Each side received the same volume of concentrated bone marrow and the same number of cells (95,000 ± 25,000 cells; counted as CFU-F).In the cadaver tests, the distance to the desired center-point of the lesion in the navigated group (1.6 mm) was significantly less than that in the control group (5.9 mm; p<0.001). Significant differences were also found in the number of drilling corrections (p<0.001), the radiation time needed (p<0.001), the risk of intra-articular penetration, and the risk of ligament injuries. In patients, computer navigation achieved results closer to the ideal position of the trocar, with better trocar placement in terms of tip-to-subchondral distance and ideal center position within the target for injection of stem cells. At the most recent follow-up (5 years), an increase in precision with computer-assisted navigation resulted in less collapse (4 vs. 1) and better volume of repair (11.4 vs 4.2 cm3) for knees treated with the computer-assisted technique. Failures were related to missing the target with intra-articular penetration.Computer-assisted navigation improved precision with less radiation. The findings of this study suggest that computer navigation may be safely used in a basic procedure for the injection of stem cells in knee osteonecrosis.