Although the rat has been the predominant rodent used to investigate the pathophysiology and treatment of experimental spinal cord injury (SCI), the increasing availability of transgenic animals has led to greater use of mouse models. However, behavioral assessment after SCI in mice has been less extensively investigated than in rats and few studies have critically examined the correlation between behavioral tests and injury severity or tissue damage. The present study characterized hindlimb functional performance in C57Bl/6 mice after contusion SCI at T9 using the weight drop method. A number of behavioral tests were examined with regard to variability, inter-rater reliability, and correlation to injury severity and white matter sparing. Mice were subjected to sham, mild-moderate or moderate-severe SCI and evaluated at day 1 and weekly up to 42 days using the Basso mouse scale (BMS), ladder climb, grid walk, inclined plane, plantar test and tail flick tests. The ladder climb and grid walk tests proved sub-optimal for use in mice, but modifications enhanced their predictive value with regard to injury severity. The inclined plane, plantar test and tail flick test showed far too much variability to have meaningful predictive value. The BMS score proved reliable, as previously reported, but a combined score (BLG) using BMS, Ladder climb (modified), and Grip walk (modified grid walk) provided better separation across injury levels and less variability than the individual tests. These data provide support for use of a combined scoring method to follow motor recovery in mice after contusion SCI.