An animal model of transected spinal cord injury (SCI) was used to test the hypothesis that cografted human umbilical mesenchymal stem cells-derived neurospheres (HUMSC-NSs) and BDNF can promote morphologic and functional recoveries of injured spinal cord. In vitro, HUMSC-NSs terminally differentiated into higher percentages of cells expressing neuronal markers: beta-tubulin III and MAP2ab by the supplement with BDNF. Following grafted into injured spinal cord, very few grafted cells survived in the HUMSC-NSs + BDNF-treated (<3%) and HUMSC-NSs-treated (<1%) groups. The survived cells were differentiated into various cells, which were confirmed by double staining of BrdU and neural or glia markers. In comparison, more grafted cells in the HUMSC-NSs + BDNF group transformed into mature neural-like cells, while more grafted cells in the HUMSC-NSs group transformed into oligodendrocyte-like cells. HUMSC-NSs + BDNF-treated group had more greatly improved BBB scores, compared with HUMSC-NSs-treated and medium-treated groups. Additionally, axonal regeneration showed significant improvement in rats receiving HUMSC-NSs + BDNF, compared with HUMSC-NSs-treated and medium-treated groups, as demonstrated by the NF-200-positive staining and Fluorogold (FG) retrograde tracing study. Lastly, a significant reduction in the percentage cavitation was seen in the two cell-treated groups compared with medium control group. These results means BDNF could promote the neural differentiation of HUMSC-NSs in vitro and in vivo. However, cellular replacement is unlikely to explain the improvement in functional outcome. The functional recovery might more rely on the axonal regeneration and neuroprotective action that active by the grafted cells. Cografted HUMSCs and BDNF is a potential therapy for SCI.