The therapeutic effect of glycosaminoglycan polysulfuric acid ester (GAGPS) was studied using the Pond-Nuki model of canine osteoarthritis. The clinical setting was simulated by permitting 4 weeks ambulation without treatment, following anterior cruciate transection. Animals were then injected with GAGPS, 4 mg/kg intramuscularly, twice weekly during weeks 4-8. Control animals received intramuscular saline. The study was terminated 4 weeks after completion of the GAGPS or saline regimen (i.e., 12 weeks postoperatively). Cartilage from the medial femoral condyle was analyzed for collagen integrity (swelling properties), hydroxyproline, uronic acid, active and total proteoglycan (PG)-degrading metalloproteinase, PG-degrading serine proteinase, and histopathology (Mankin score). Condylar cartilage from animals treated with GAGPS demonstrated less cartilage swelling, less total and active metalloproteinase, and lower histopathologic scores than were found in cartilage from saline-treated animals. GAGPS was able to suppress PG-degrading enzyme activity and maintain a more normal-appearing cartilage. It is proposed that GAGPS suppressed PG breakdown by decreasing synthesis of metalloproteinase or by directly inhibiting metalloproteinase in cartilage, rather than by increasing synthesis of PG by chondrocytes.