Mesenchymal stem cells (MSCs) have been used to repair connective tissue defects in several animal models. Compared to "natural healing" controls (no added cells), MSC-collagen gel constructs in rabbit tendon defects significantly improve repair biomechanics. However, ectopic bone forms in 28% of MSC-treated rabbit tendons. To understand the source of bone formation, three studies were performed. In the first study, the hypothesis was tested that MSCs delivered during surgery contribute to bone formation in the in vivo repair site. Adjacent histological sections in the MSC-treated repair tissue were examined for pre-labeled MSCs and for cells showing positive alkaline phosphatase (ALP) activity. Both cells were observed in serial sections in regions of ectopic bone. Contralateral "natural healing" tendons lacked both markers. In the other two studies, the effects of osteogenic supplements and construct geometry (monolayer vs. 3-D) on ALP activity were studied to test three hypotheses: that rabbit MSCs increase ALP activity over time in monolayer culture conditions; that adding osteogenic inducing supplements to the culture medium increases cellular protein in monolayer culture; and that rabbit MSCs increase ALP activity both in monolayer and in 3-D constructs, with and without media supplements. Culture in monolayer under similar conditions to in vivo (as in the first study) did not increase ALP at 2 or 4 weeks. Medium designed to increase osteogenic activity significantly increased cell numbers (cellular protein increased by 260%) but did not affect ALP activity either in monolayer or 3-D constructs (p>0.12). However, MSCs in 3-D constructs exhibited higher ALP activity than cells in monolayer, both in the presence (p<0.045) and absence of supplement (p<0.005). These results suggest that in vitro conditions may critically influence cell differentiation and protein expression. Mechanisms responsible for these effects are currently under investigation.