The 'Allen technique', a weight-drop procedure introduced in 1911, remains the most widely used technique for experimental spinal cord contusion. Control of injury severity in this procedure is achieved by alteration of the height from which the standardized weight (usually 20 g) is dropped. It has not been possible in this technique to independently vary the amount of cord compression and the initial velocity of compression, since both are related to drop-weight kinetic energy. Our approach uses a constrained stroke pneumatic impactor to afford independent control of these two parameters. Mechanical testing of the device has verified the accuracy and repeatability of impact velocity and cord compression. More importantly, a pilot series at 2 m/s contact velocity with a range of compression has demonstrated neurophysiologically distinct levels of spinal cord injury as a function of compression. This includes a 'moderate' functional injury with impaired and delayed neuronal conduction through the injury site. Such a 'moderate' injury, at the threshold between recovery and permanent cord dysfunction, is particularly promising for the study of mechanisms underlying progressive post-contusion pathology; 'moderate' injury has not been reproducibly generated in weight-drop techniques.