Arthroscopy allows direct visual examination of joint cavity components and is useful for the diagnosis, treatment and evaluation of lesions. We investigated the contribution of arthroscopy to the evaluation of joint cartilage. The severity of cartilage lesions can be assessed using a total 100-mm visual analog scale (0 = no chondropathy; 100 = the worst possible lesions) or a more objective system based on the site, depth, and surface area of the lesions. This latter system was developed by the French Society for Arthroscopy (Société Française d'Arthroscopie) and provides a score and a class (SFA score and SFA grade). We investigated whether this system has the characteristics required of an evaluation tool, i.e., simplicity, reproducibility, clinical relevance, sensitivity to change, and discriminant capacity. Arthroscopy is an invasive procedure. However, we introduced several simplifications, including use of local rather than general anesthesia, performance on an outpatient basis, elimination of the tourniquet (to avoid muscular dysfunction), and use of a small arthroscope. This simplified technique is called chondroscopy. Intra-observer reproducibility is far better than inter-observer reproducibility. We found a good correlation between the two arthroscopy scales (visual analog scale and SFA scale). Chondroscopy and roentgenographic evaluations of cartilage lesions were closely correlated. Changes in the severity of cartilage lesions were correlated with changes in functional impairment. Chondroscopy proved capable of demonstrating statistically significant changes in cartilage lesions due to knee osteoarthritis between two evaluations done only one year apart, even in a small sample of patients (less than 20). A preliminary study of repeated hyaluronic acid injections suggested that chondroscopy may be capable of identifying truly chondromodulating agents.