Controversies exist concerning the need for mesenchymal stromal cells (MSCs) to be transdifferentiated prior to their transplantation. In the present study, we compared the results of grafting into the rat contused spinal cord undifferentiated, adipose tissue-derived stromal cells (uADSCs) versus ADSCs induced by two different protocols to form differentiated nervous tissue. Using Basso, Beattie, and Bresnahan scores and grid tests, we found that three cell-treated groups, including uADSCs-treated, dADSCs induced by Protocol 1 (dADSC-P1)-treated, and dADSCs induced by Protocol 2 (dADSC-P2)-treated groups, significantly improved locomotor functional recovery in SCI rats, compared with the saline-treated group. Furthermore, functional recovery was better in the uADSC-treated and dADSC-P2-treated groups than in the dADSC-P1-treated group at week 12 postinjury (P < 0.05 for dADSC-P1 group vs. uADSCs or dADSC-P2 groups). Although both protocols could induce high percentages of cells expressing neural markers in vitro, few BrdU-labeled cells survived at the injury sites in the three cell-treated groups, and only a small percentage of BrdU-positive cells expressed neural markers. On the other hand, the number of NF200-positive axons in the uADSC-treated and dADSC-P2-treated groups was significantly larger than those in the dADSC-P1-treated and saline-treated control groups. Our results indicate that ADSCs are able to differentiate into neural-like cells in vitro and in vivo. However, neural differentiated ADSCs did not result in better functional recovery than undifferentiated ones, following SCI. In vitro neural transdifferentiation of ADSCs might therefore not be a necessary pretransplantation step. Furthermore, cellular replacement or integration might not contribute to the functional recovery of the injured spinal cord.