45 orthopaedic residents were invited to participate in this study evaluating improvement in knee arthroplasty performance, and were randomized to either cadaveric training, a virtual reality simulation, or no training. The aim of this study was to determine the efficacy of each training modality in improving knee arthroscopy performance among residents, and to assess the cost-effectiveness of each method. Video recordings were taken of pre- and post-training performance tests in order to assess the Arthroscopic Surgical Skills Evaluation Tool (ASSET), as was the time to completion of the test. Results indicated significant performance improvement among the cadaveric training group, non-significant improvement in the virtual reality training group, and no improvement in the control group participants' performance. However, cadaveric training was also the most costly option between cadaveric and virtual reality training, and cost-effectiveness per ASSET point improvement was estimated to be similar between cadaveric and simulator training.