This study evaluates the ability of MSCs isolated from different origins--bone marrow, periosteum, or fat--to treat partial growth arrest in immature (6-week-old) New Zealand White rabbits. Up to 50% of the medial half of the proximal physis of the tibia was excised in these New Zealand White rabbits. Three weeks later, the bony bridge was excised, and fibrin glue with and without MSCs were transferred into the physeal defect of different rabbits. Contralateral tibias, without undergoing operation, served as self-control. Four groups of rabbits were involved in the study. Each group was injected separately with bone marrow-derived MSCs (group I), periosteum-derived MSCs (group II), fat-derived MSCs (group III), and fibrin glue alone (control, group IV). The rabbits were killed 8 and 16 weeks postoperatively. Clinical, radiological, and histological analyses were subsequently performed. Similar proliferative rates for three MSCs were demonstrated on days 4, 7, and 11 of primary culture. However, MSCs derived from bone-marrow and periosteum appeared to be more homogeneous than that from fat. All MSCs demonstrated chondrogenic and osteogenic differentiation potentials in vitro. The tibias in groups I and II showed significant correction of varus angulation at 16 weeks. However, the varus angulation in group III remained significantly obvious when compared with group I (p < 0.05). The length discrepancies between operated and normal tibiae in groups I, II, and III were significantly corrected compared with control (p < 0.01). In conclusion, bone-marrow and periosteum yielded more homogenous MSCs than fat, providing better correction of physeal arrest in rabbits. The source of MSCs itself could influence the success in the treatment of growth arrest.