Spinal cord injury (SCI) may cause disastrous damage to human locomotion and ultimately make patients suffer from gait anomaly. In the extensive SCI research, the locomotion function serves as a vital standard not only for revealing the underlying SCI mechanism but also for evaluating the clinical therapy. Gait division is the basis of gait analysis. Calculation of gait parameters is available for locomotion function evaluation only when gait cycles are accurately divided. Based on the characteristics of stride height, which is defined as the real-time height of toes vertical to the running direction of a treadmill belt, this study presented three automatic gait division methods, divided the gait cycles for healthy and spinal cord-injured rhesus monkeys, established the evaluation standards, and made comparison of these three methods. For the healthy, injured and mixed groups, the overall accuracies of these three methods were respectively 0.871[Formula: see text][Formula: see text][Formula: see text]0.223, 0.570[Formula: see text][Formula: see text][Formula: see text]0.372, and 0.720[Formula: see text][Formula: see text][Formula: see text]0.339 (method 1); 0.658[Formula: see text][Formula: see text][Formula: see text]0.245, 0.737[Formula: see text][Formula: see text][Formula: see text]0.206, and 0.698[Formula: see text][Formula: see text][Formula: see text]0.228 (method 2); 0.966[Formula: see text][Formula: see text][Formula: see text]0.060, 0.759[Formula: see text][Formula: see text][Formula: see text]0.343, and 0.863[Formula: see text][Formula: see text][Formula: see text]0.265 (method 3). The results show that the stride height characteristics combined with the filter technique may help realize the adequate gait division.