The purpose of this study was to develop a minimally invasive recovery model of spinal cord injury in the C57Bl/6J mouse. Without laminectomy, the epidural space was exposed by disruption of the T10-T11 interspinous ligament. Perpendicular to the rostral-caudal axis of the spine, a 1.5-mm silicone tube (O.D. 0.047 in.) was placed in the T11 epidural space. Prior to placement, a suture was passed through the tube allowing withdrawal of the tube after discontinuation of anesthesia. After 1, 30, 60, or 120 min (n = 5-8) of spinal cord compression (SCC), the tube was withdrawn. Neurological function was measured at 1, 3, 7, and 14 days after injury followed by histologic analysis. BBB locomotor score, rotarod latency, and screen grasping were worsened in a SCC duration-dependent manner (p < 0.0001). With increasing SCC duration, the number of histologically normal neurons in the ventral horns decreased (p < 0.0001) while the cross-sectional area of spinal cord with pancellular necrosis increased (p < 0.0001). Increased duration of SCC caused progressive rostral-caudal spread of histologic damage. The results indicate that this is a simple, reliable model with neurologic and histologic injury highly dependent on SCC duration. This model may be useful for study of spinal cord injury in genetically modified mice in the absence of anesthetic confounds while leaving the vertebral column intact.