As information on the in vivo niche of tendon stem cells would benefit the mobilization and ex vivo culture of tendon stem cells for tendon repair, I will summarize the possible strategies that might be useful for tracking the fate of tendon stem cells in vivo. In vitro MSC markers could be used to locate positive cells in vivo using immunohistochemistry or in situ hybridization. This approach, although sensitive, is currently limited by the lack of specific MSC or tendon stem cell markers. Some of the MSC markers such as CD44, CD90, CD73, CD29 and CD105 were not specific to MSCs and were also expressed by fibroblasts [77, 78]. Another approach is to inject labelled cultured stem cells into the circulation to analyse their tissue distribution in vivo. This strategy might be less accurate to study the natural distribution of tendon stem cell in vivo because the cells might engraft non-specifically in different organs and specific homing signals might be required for recruiting the injected stem cells to tendons. Tendon has a poor blood supply and this might also affect the recruitment of injected cells to healthy tendon. The injection of cultured stem cells directly into an injured tendon is possible, but may not be feasible in an intact tendon with tightly packed and organized collagen fibres. How injection-induced tendon injury may affect the results remains unclear. Dudhia et al.  compared the amount of labelled BMSCs in tendon lesions of horses with tendinopathies or desmopathies using intralesional, intravenous and regional perfusion routes. They showed that intralesional administration of BMSCs retained the highest number of cells, followed by regional perfusion. Intravenous injection of BMSCs resulted in distribution of cells largely to the lung fields and there were no detectable cells in the tendon lesions . Exogenous injection of tendon cells hence might be less accurate in studying the natural distribution and functions of tendon stem cells and it might not reflect or might even disturb the endogenous tendon stem cell activities. The use of bromodeoxyuridine (BrdU) labelling to identify label-retaining cells (LRC) with long cell-cycle time or asymmetric-cell division with non-random chromosomal cosegregation theoretically is useful for the localization of tendon stem cells in vivo as they are supposed to be quiescent and retain the label while the differentiated cells proliferate and lose the BrdU signal rapidly during the washout period . Using a double nucleoside analogue cell-labelling system (IdU/CldU), Kurth et al.  reported the identification of a population of quiescent, slow-cycling, non-hematopoietic, non-endothelial, MSC-like stromal cells, present in both the lining layer and sublining tissue of synovium of knee joint in vivo. However, this method is not specific for stem cells and might label cells that have stopped proliferating because of various reasons (e.g. differentiation) and hence might be subjected to false-positive errors. The self-renewal capacity of stem cells was suggested to correlate with telomerase activity . Based on this hallmark of stem cells, Breault et al.  have generated mTert-GFP-transgenic mice as a model system to mark male germ cells, hematopoietic stem cells (HSCs) and intestinal crypt cells (ISCs) in vivo. The feasibility of using this system to mark stem cells in tendon needs further research. Similar to the BrdU labelling method, this method also does not specifically label resident stem cells in tendons and hence the relative contribution of stem cells from different sources to tendon healing or failed healing cannot be revealed and requires the combined use of tissue-specific markers to elucidate the mechanism. To look systematically for niche in tendon tissue, the ideal method is to mark the tendon stem cells using genetic-based lineage tracing technique and follow their lineages. Information related to tissue development is usually taken into consideration in the selection of appropriate markers for lineage tracing and hence this method is more specific. Using the same approach, Feng et al.  have used NG2-driven Cre to trace pericytes. Besides, Lounev et al.  have used MyoD-Cre, Tie2-Cre and smooth muscle myosin heavy chain-Cre (SMMHC-Cre) to trace the possible involvement of skeletal muscle stem cells, endothelial precursors and vascular smooth muscle cells, respectively, in heterotopic muscle ossification . Speer et al.  used SM22-Cre to genetically trace cells derived from smooth muscle and found that smooth muscle cells gave rise to osteochondrogenic precursor- and chondrocyte-like cells in calcified blood vessels of matrix Gla protein deficient (MGP−/−) mice. Recently, tenomodulin, scleraxis and thrombospondin 4 have been suggested to be more specific biomarkers for tendon fibroblasts and were discussed in a review article . Whether these tendon-related markers could be used for tendon lineage tracing and hence for understanding the in vivo identity and roles of tendon stem cells needs further experiments. Table 1 summarizes the possible strategies, their advantages and limitations, for tendon stem cell tracking in vivo.