Most previous studies on spinal cord injury (SCI) have used rodent models. Direct extrapolation of the results obtained in rodents to clinical cases is difficult, however, because of neurofunctional and anatomic differences between rodents and primates. In the present study, the development of histopathologic changes and functional deficits were assessed quantitatively after mild, moderate, and severe spinal cord contusive injuries in common marmosets. Contusive SCI was induced by dropping one of three different weights (15, 17, or 20 g) at the C5 level from a height of 50 mm. Serial magnetic resonance images showed significant differences in the intramedullary T1 low signal and T2 high signal areas among the three groups. Quantitative histologic analyses revealed that the number of motor neurons, the myelinated areas, and the amounts of corticospinal tract fibers decreased significantly as the injury increased in severity. Motor functions were evaluated using the following tests: original behavioral scoring scale, measurements of spontaneous motor activity, bar grip test, and cage-climbing test. Significant differences in all test results were observed among the three groups. Spontaneous motor activities at 10 weeks after injury were closely correlated with the residual myelinated area at the lesion epicenter. The establishment of a reliable nonhuman primate model for SCI with objective functional evaluation methods should become an essential tool for future SCI treatment studies. Quantitative behavioral and histopathologic analyses enabled three distinct grades of injury severity (15-g, 17-g, and 20-g groups) to be characterized with heavier weights producing more serious injuries, and relatively constant behavioral and histopathologic outcomes. © 2005 Wiley-Liss, Inc.