Locomotor training on a treadmill is a therapeutic strategy used for several years in human paraplegics in whom it was shown to improve functional recovery mainly after incomplete spinal cord lesions. The precise mechanisms underlying its effects are not known. Experimental studies in adult animals were chiefly performed after complete spinal transections. The objective of this experiment was to assess the effects of early treadmill training on recovery of spontaneous walking capacity after a partial spinal cord lesion in adult rats. Following a compression-injury by a subdurally inflated microballoon, seven rats were trained daily on a treadmill with a body weight support system, whereas six other animals were used as controls and only handled. Spontaneous walking ability in an open field was compared weekly between both groups by two blinded observers, using the Basso, Beattie and Bresnahan (BBB) locomotor rating scale. Mean BBB score during 12 weeks was globally significantly greater in the treadmill-trained animals than in the control group, the benefit of training appearing as early as the 2nd week. At week 7, locomotor recovery reached a plateau in both animal groups, but remained superior in trained rats. Daily treadmill training started early after a partial spinal cord lesion in adult rats, which accelerates recovery of locomotion and produces a long-term benefit. These findings in an animal model mimicking the closed spinal cord injury occurring in most human paraplegics are useful for future studies of optimal locomotor training programs, their neurobiologic mechanisms, and their combination with other treatment strategies.