Abstract Study question This study aims to compare the engraftment, retaining time and therapeutic efficiency of differently administered BMSCs and help to select an optimal therapeutic route in clinical settings. Summary answer Compared with intrauterine infusion, BMSCs could better promote angiogenesis by upregulating related cytokines, such as VEGF, when administered through the ipsilateral iliac artery. What is known already MSC-based therapy has become a promising method for endometrial disease(thin endomtrium or Ashernmen’s syndrowe). Therapeutic effects could always be observed even though different MSC administration routes or MSCs of different tissue sources were used in these studies. Only a few studies compared efficacy of different transplantation routes. However, the results seem to be controversial. Comparable therapeutic effects were reported in some studies, while others stated that systematic administration gave a better outcome than local administration. Study design, size, duration Experimental animal study. Forty-eight female Sprague-Dawley (SD) rats were used in this study. They were randomly assigned to 4 groups: normal, injured, intra-arterial and intra-uterine group. For all rats except for normal group, the thin endometrium models were established by infusing 95% ethanol into the uterine horns and BMSCs were transplanted either locally or intra-arterially after modeling. The therapeutic efficacy were evaluated in the following month. Participants/materials, setting, methods The thin endometrium models induced by ethanol in SD rats, GFP/Luciferin labeled BMSCs were injected either locally or intra-arterially. The retaining time and quantitative distribution were assessed by in vivo bioluminescence imaging and immune-histological analysis. The precise location and differentiation of differently administered BMSCs were determined by immunofluorescence methods. The endometrial fibrosis, angiogenesis were detected by immunohistochemistry and western blotting at a consecutive time after treatment to compare the therapeutic efficiency of two administration methods. Main results and the role of chance The engraftment and differentiation abiility were comparable in 2 groups. The luminescent signal both remained distinct and strong in the abdomen in the first 4 days post-treatment(7.98 × 105 and 6.02 × 105p/s for IU and IA group), indicating the precise and concentrated distribution of BMSCs administered both locally or intra-arterially. The luminescent signals disappeared under bioluminescence imaging over time. We further evaluated the precise distribution, differentiation ability and retaining time of the BMSCs delivered in two strategies by immunofluorescence analysis. All the GFP positive cell localized in stroma, but not in the epithelium or myometrium. Furthermore, there are significantly more positive staining in basal layer of the endometrium close to the glands and vessels than the outer layer of the endometrium in the intra-arterial group. At the 28th days post treatment, we could capture a few GFP staining in the basalis layel of endometrium in intra-arterial group and there were no GFP fluorescence signals detected in intra-uterine group(P < 0.05), suggesting a better survival of BMSCs administered intra-arterially. Differentiation ability of differently administered BMSCs were similar. A few BMSCs began to differentiate into stromal cell 12 days after therapy. Limitations, reasons for caution No pregnancy tests were carried out in these rats to further confirm the regeneration of thin endometrium and compare the therapeutic efficacy. Wider implications of the findings: Our study unveiled that the location of MSCs might determined their regenerarive ability and retaining time, and provided an optimal therapeutic route in clinical settings. Trial registration number Not applicable