Objective To investigate tissue engineered spinal cord which was constructed of bone marrow mesenchymal stem cells (BMSCs) seeded on the chitosan-alginate scaffolds bridging the both stumps of hemi-transection spinal cord injury (SCI) in rats to repair the acute SCI. Methods BMSCs were separated and cultured from adult male SD rat. Chitosan-alginate scaffold was produced via freeze drying, of which the structure was observed by scanning electron microscope (SEM) and the toxicity was determined through leaching l iquor test. Tissue engineered spinal cord was constructed by seeding second passage BMSCs on the chitosan-alginate scaffolds (1 × 106/mL) in vitro and its biocompatibil ity was observed under SEM at 1, 3, and 5 days. Moreover, 40 adult female SD rats were made SCI models by hemi-transecting at T9 level, and were randomly divided into 4 groups (each group, n=10). Tissue engineered spinal cord or chitosan-alginate scaffolds or BMSCs were implanted in groups A, B, and C, respectively. Group D was blank control whose spinal dura mater was sutured directly. After 1, 2, 4, and 6 weeks of surgery, the functional recovery of the hindl imbs was evaluated by the Basso-Beattie-Bresnahan (BBB) locomotor rating score. Other indexes were tested by wheat germ agglutinin-horseradish peroxidase (WGA-HRP) retrograde tracing, HE staining and immunofluorescence staining after 6 weeks of surgery. Results Chitosan-alginate scaffold showed three-dimensional porous sponge structure under SEM. The cells adhered to and grew on the surface of scaffold, arranging in a directional manner after 3 days of co-culture. The cytotoxicity of chitosan-alginate scaffold was in grade 0-1. At 2, 4, and 6 weeks after operation, the BBB score was higher in group A than in other groups and was lower in group D than in other groups; showing significant differences (P lt; 0.05). At 4 and 6 weeks, the BBB score was higher in group B than in group C (P lt; 0.05). After 6 weeks of operation, WGA-HRP retrograde tracing indicated that there was no regenerated nerve fiber through the both stumps of SCI in each group. HE and immunofluorescence staining revealed that host spinal cord and tissue engineering spinal cord l inked much compactly, no scar tissue grew, and a large number of neurofilament 200 (NF-200) positive fibers and neuron specitic enolase (NSE) positive cells were detected in the lesioned area in group A. In group B, a small quantity of scar tissue intruded into non-degradative chitosan-alginate scaffold at the lesion area edge, and a few of NSE flourescence or NF-200 flourescence was observed at the junctional zone. The both stumps of SCI in group C or group D were filled with a large number of scar tissue, and NSE positive cells or NF-200 positive cells were not detected. Otherwise, there were obviously porosis at the SCI of group D. Conclusion The tissue engineered spinal cord constructed by multi-channel chitosan-alginate bioscaffolds and BMSCs would repair the acute SCI of rat. It would be widely appl ied as the matrix material in the future.