OBJECTIVE: To evaluate the effects of moderate exercise on glycosaminoglycan (GAG) content in knee cartilage in subjects at high risk of knee osteoarthritis (OA). METHODS: Forty-five subjects (16 women, mean age 46 years, mean body mass index 26.6 kg/m(2)) who underwent partial medial meniscus resection 3-5 years previously were randomized to undergo a regimen of supervised exercise 3 times weekly for 4 months or to a nonintervention control group. Cartilage GAG content, an important aspect of the biomechanical properties of cartilage, was estimated by delayed gadolinium-enhanced magnetic resonance imaging of cartilage (dGEMRIC), with results expressed as the change in the T1 relaxation time in the presence of Gd-DTPA (T1[Gd]). RESULTS: Thirty of 45 patients were examined by dGEMRIC at baseline and followup. The exercise group (n = 16) showed an improvement in the T1(Gd) compared with the control group (n = 14) (15 msec versus -15 msec; P = 0.036). To study the dose response, change in the T1(Gd) was assessed for correlation with self-reported change in physical activity level, and a strong correlation was found in the exercise group (n = 16, r(S) = 0.70, 95% confidence interval [95% CI] 0.31-0.89) and in the pooled group of all subjects (n = 30, r(S) = 0.74, 95% CI 0.52-0.87). CONCLUSION: This in vivo cartilage monitoring study in patients at risk of knee OA who begin exercising indicates that adult human articular cartilage has a potential to adapt to loading change. Moderate exercise may be a good treatment not only to improve joint symptoms and function, but also to improve the knee cartilage GAG content in patients at high risk of developing OA.