OBJECTIVE To explore an accurate neurophysiological technique that demonstrates small functional differences after spinal cord injury and assesses therapeutic interventions. METHODS A modified weight drop (WD) technique was used at T8 in rats to build graded spinal cord injury model. Rubrospinal MEPs were recorded at T13 epidurally to monitor spinal cord function in end week 4 after graded spinal cord injury. The efficacy of this techniques to monitor spinal cord function was compared to BBB locomotor rating scale and histologic evaluation. RESULTS A characteristic peak complex of rubrospinal MEPs in sham-operated group consisted of 5-7 positive waves and 4-5 negative waves emerging after red nucleus stimulation. The summed peak to peak amplitude (for practical reasons, called peak amplitude) was (195.25 +/- 34.35) microV and decreased following spinal cord injury. The latency of the first peak (positive wave) was (1.57 +/- 0.15) ms and prolonged following spinal cord injury. Significant Linear relationship existed between the peak amplitude and the BBB scores (r = 0.79) and between the peak amplitude and the residual matter obtained from the section with maximum tissue damage( r = 0.87). The close relationship between the latency of the first peak and the BBB scores (r = -0.88) and between the latency of the first peak and residual matter (r = -0.86) were observed. CONCLUSIONS Amplitudes and latencies of rubrospinal MEPs are very valuable parameters to demonstrate small function differences. Rubrospinal MEPs can be used as a reliable measure for motor function prognosis after spinal cord injury.