Mesenchymal stem cell (MSC) therapy has emerged as a potentially powerful tool for osteochondral repair, but the safety of stem cells in vivo is still not clear. In this study, we assessed the safety and efficacy of canine adipose tissue-derived MSCs (cATMSCs) or umbilical cord blood-derived MSCs (cUCBMSCs) in osteochondral repair. The test cells were implanted at the site of an osteochondral defect surgically created on the trochlear groove of beagle dogs. To address safety concerns in this model, we conducted a GLP-compliant study. We observed no significant treatment-related changes in body weights, clinical signs, food consumption, hematological or biochemical values, organ weights, or histopathological findings. The stem cell implants induced no detectable tumor formation or local or systemic rejection reaction. In the assessment of effectiveness, 13-week samples from both the cATMSC- and cUCBMSC-implanted groups showed fibrous tissue formation in the extracellular matrix scaffold of the defect similar to that of controls. However, repaired tissues in stem cell-seeded groups showed partial differentiation toward an osteochondral phenotype. These findings support the safety of cATMSC or cUCBMSC transplantation, and show that the cells may potentially differentiate into osteochondral tissues in our canine model. Our study provides a framework in which to establish preclinical safety standards for the development of cellular therapeutics.