The purpose of this study was to evaluate the performance (simplicity, reproducibility, relevance) of chondroscopy as a method for evaluating cartilage damage. Chondroscopy consisted in endoscopic evaluation of the knee using a 2.7-mm Storz arthroscope under local anesthesia and recorded on videotape. Scoring of chondropathy was based on physician's overall assessment using a 100-mm-length visual analogue scale (VAS) and size and grade of cartilage lesions. Reproducibility was evaluated by variability (coefficient of variation [CV]) in the reading of chondroscopic evaluations of five patients five times by one physician and one time each by four different physicians. The correlations between scoring of chondropathy (VAS) and radiological articular joint space narrowing, demographic data (sex, age, weight), and disease characteristics (localization, etiology, activity) were studied in 84 outpatients fulfilling the American College of Rheumatology criteria for the diagnosis of osteoarthritis of the knee. The grade and size of the lesions were both correlated with the physician's overall assessment (r = 0.713 and r = 0.816, respectively). These two variables accounted for 72% of the variance of the VAS (multiple regression analysis). Intraobserver reproducibility was better than interobserver reproducibility (CV, 9% and 37%, respectively). There was a strong correlation between the scoring of chondropathy (VAS) and radiological joint space narrowing (r = .646, P < .0001). Moreover, in 17 of 33 patients without radiological joint space narrowing, VAS was > 20 mm. At variance, the body mass index was the single clinical variable found to correlate with the scoring of chondropathy (r = .282, P < .001).(ABSTRACT TRUNCATED AT 250 WORDS)