Background The use and effectiveness of hematopoietic stem cell transplantation (HSCT) are limited by lethal complications, i.e., acute and chronic graft-versus-host disease (aGVHD and cGVHD, respectively), in which immune cells from the donor attack healthy recipient tissues. GVHD presents both prophylactic and therapeutic challenges, and overall survival is poor. Mesenchymal stem cells (MSCs) show considerable promise in the treatment of GVHD because of their potential immunomodulatory activity. Multiple studies have been performed to explore the possible benefit of MSCs in GVHD, but the results of these studies are sometimes conflicting. Therefore, we performed a systematic review and meta-analysis to estimate the effect of MSC infusion on GVHD treatment and prevention. Methods We systematically searched the MEDLINE (PubMed), Cochrane Library, EMBASE, ClinicalTrials.gov, and SinoMed CBM databases to identify studies published before February 2018 involving patients with hematologic malignancies undergoing HSCT and receiving MSC-based or conventional therapy. We included studies if they reported on the outcomes of interest. Results Ultimately, 10 studies were selected from among 413 candidates. According to our meta-analyses, compared with conventional treatment, MSC therapy demonstrated substantial improvements in terms of complete response (CR) and overall survival for cGVHD. However, MSC therapy did not show substantial improvements in terms of engraftment, the incidence of aGVHD, relapse, death, death due to relapse, or death due to infection. Subgroup analyses showed that MSCs derived from the umbilical cord (U-MSCs) and MSC infusion after HSCT substantially improved engraftment and cGVHD incidence, whereas MSCs derived from bone marrow (B-MSCs) and MSC infusion before HSCT shows no improvement. In addition, B-MSCs and MSC infusion before HSCT tend to prolong engraftment time, as well as increase the rates of relapse and death. Conclusions MSC infusion can reduce cGVHD but not aGVHD incidence and showed a positive effect in patients who already had aGVHD. For GVHD prevention, the use of U-MSCs and MSC infusion after HSCT were optimal for reducing cGVHD incidence and promoting engraftment, and might help decrease the incidence rate of relapse and death. However, B-MSCs and MSC infusion before HSCT may be harmful to patients and thus require serious consideration. A lack of robust evidence, owing to the small number of studies and small sample sizes, indicates a need for further high-quality clinical trials including large numbers of patients to validate our findings. Electronic supplementary material The online version of this article (10.1186/s13287-019-1287-9) contains supplementary material, which is available to authorized users.