Background Contractures are a prevalent and potentially severe complication in patients with neurologic disorders. Although heat, cold, and stretching are commonly used for treatment of contractures and/or spasticity (the cause of many contractures), the sequential effects of these modalities remain unclear. Questions/purposes Using an established rat model with spinal cord injury with knee flexion contracture, we sought to determine what combination of heat or cold before stretching is the most effective for treatment of contractures derived from spastic paralyses and investigated which treatment leads to the best (1) improvement in the loss of ROM; (2) restoration of deterioration in the muscular and articular factors responsible for contractures; and (3) amelioration of histopathologic features such as muscular fibrosis in biceps femoris and shortening of the joint capsule. Methods Forty-two adolescent male Wistar rats were used. After spasticity developed at 2 weeks postinjury, each animal with spinal cord injury underwent the treatment protocol daily for 1 week. Knee extension ROM was measured with a goniometer by two examiners blinded to each other's scores. The muscular and articular factors contributing to contractures were calculated by measuring ROM before and after the myotomies. We quantitatively measured the muscular fibrosis and the synovial intima length, and observed the distribution of collagen of skeletal muscle. The results were confirmed by a blinded observer. Results The ROM of heat alone (34°± 1°) and cold alone (34°± 2°) rats were not different with the numbers available from that of rats with spinal cord injury (35°± 2°) (p = 0.92 and 0.89, respectively). Stretching after heat (24°± 1°) was more effective than stretching alone (27°± 3°) at increasing ROM (p \ 0.001). Contrastingly, there was no difference between stretching after cold (25°± 1°) and stretching alone (p = 0.352). Stretching after heat was the most effective for percentage improvement of muscular (29%) and articular (50%) factors of contractures. Although quantification of muscular fibrosis in the rats with spinal cord injury (11% ± 1%) was higher than that of controls (9% ± 0.4%) (p = 0.01), no difference was found between spinal cord injury and each treatment protocol. The total synovial intima length of rats with spinal cord injury (5.9 ± 0.2 mm) became shorter than those of the controls (7.6 ± 0.2 mm) (p \ 0.001), and those of stretching alone (6.9 ± 0.4 mm), stretching after heat (7.1 ± 0.3 mm), and stretching after cold (6.7 ± 0.4 mm)