A detailed examination of the histopathological features of the clip compression injury in mice was performed to understand the relationships between neurological function and existing pathology of the spinal cord. Adult, female CD1 mice underwent three grades of extradural clip compression injury (3-g, 8-g, and 24-g FEJOTA mouse clips), transection, and sham injury at T3-4. Quantitative behavioural assessments were performed for 4 weeks following SCI. After 4 weeks, Fluoro-Gold was introduced caudal to the SCI site, at T9, and was retrogradely transported for 5 days to the origin of spared axons through the injury site. Counts of retrogradely labeled neurons in the brain-stem, midbrain, and sensory-motor cortex indicated that the number of intact descending axons that traversed the lesion decreased with increasing injury severity (F > 28; df = 4; p < 0.0001; one-way ANOVA). Independent linear correlation analyses were performed between indices of neurological recovery (BBB and IP test), counts of retrogradely labeled neurons and morphometric assessments of normal residual tissue at the injury epicenter. The BBB test correlated strongly with the amount of residual tissue at the injury epicenter (R = 0.945, df = 28, p < 0.0001). Counts of neurons retrogradely labeled with Fluoro-Gold were also strongly correlated with the BBB scores. The extrapyramidal (raphespinal, reticulospinal, vestibulospinal, and rubrospinal) tracts had Pearson correlation coefficients (R) of 0.814, 0.812, 0.813, and 0.747, respectively (df = 28, p < 0.0001). The pyramidal (corticospinal) tract had a correlation of R = 0.747, df = 28, p < 0.0001 with the BBB scores. The IP scores also correlated strongly with the persistence of extrapyramidal (raphespinal, reticulospinal, vestibulospinal and rubrospinal) tracts with correlation coefficients of 0.801, 0.782, 0.790, and 0.836, respectively (df = 28, p < 0.0001). These data indicate that the counts of retrogradely labeled neurons at the origin of distinct descending motor pathways are predictors of the variance of the functional recovery measured by the BBB and IP tests following spinal cord injury. In addition, we provide a detailed neuroanatomical study of clip compression injury in mice that can be used to study the molecular mechanisms of SCI in knockout and transgenic mice.