Outcome measures for functional assessment in experimental or naturally occurring spinal cord injury (SCI) in dogs have been largely subjective. This study is the first step in developing an easy, accurate, and objective outcome measure for neurologic dogs. The hypothesis was that the coefficient of variation (CV) of spatiotemporal parameters of gait in dogs with hindlimb paresis would be greater than that of normal dogs and dogs with orthopedic disease. This study evaluates the accuracy, sensitivity, and specificity of spatiotemporal parameters in dogs with naturally occurring SCI. All dogs were allowed to walk at their own pace over a pressure walkway. Stride time, stance time, swing time, and stride length, and velocity were recorded using the pressure walkway, and age, breed, weight, and group were recorded for each dog. The gait parameters were summarized for each dog with coefficients of variation (CsV), determined three ways. The data were analyzed with competing models to determine the best one for differentiating neurologic dogs from non-neurologic dogs. Velocity, acceleration, height, and weight did not significantly affect any of the CsV. The model with the highest accuracy (89%) was a multivariate model using the CsV (calculated by combining feet of each dog) of stride length, stride time, and swing time (p = 0.0001). The sensitivity (0.8) and specificity (0.9) were calculated using Youden's Index. The combination of CsV (combined feet) of stride length, stride time, and swing time are relatively simple and accurate with great potential as an outcome measure in dogs with SCI.