To compare the efficacy and safety of umbilical cord blood mononuclear cells (CBMNC) and azathioprine (AZA) in the treatment of patients with steroid-dependent or -resistant ulcerative colitis. One hundred and six patients diagnosed with steroiddependent or -resistant ulcerative colitis were studied retrospectively, including 36 patients treated with CBMNC and 70 treated with AZA. To reduce confounding bias due to retrospective nature of this study, the propensity score matching system was applied to equipoise the pretreatment data of two groups. After matching, 35 matched pairs (1:1) were created. The ratios of clinical remission, clinical response and endoscopic mucosal healing, Mayo score, and major complications were compared between two groups at weeks 8, 16, and 36 after treatment. The results demonstrated that the ratios of clinical remission (80% vs. 57%, P < 0.05) and mucosal healing (74% vs. 51%, P < 0.05) were significantly higher in CBMNC-treated patients compared with those in AZA-treated patients at week 8. The erythrocyte sedimentation rate was significantly decreased in CBMNC group compared with that in AZA-treated group (14.5 ± 3.9 mm/h vs. 18.0 ± 5.7 mm/h, P < 0.01) at week 8. In AZA group, 2 patients had neutropenia and 3 patients had elevated alanine aminotransferase levels, whereas no obvious side-effects were observed in CBMNCtreated group. Our results reveal that CBMNC therapy appears to be an effective and safe strategy for patients with steroid-dependent or -resistant ulcerative colitis. Further prospective studies are needed to define the potential roles and mechanisms of CBMNC in the treatment of refractory ulcerative colitis.