It is unknown whether activity-based physical therapy (ABPT) modalities that mobilize the paralyzed limbs improve bone integrity at the highly fracture-prone epiphyseal regions of the distal femur and proximal tibia following severe spinal cord injury (SCI). In this study, 4-mo-old skeletally mature littermate-matched male Sprague-Dawley rats received either SHAM surgery or severe contusion SCI. At 1 wk postsurgery, SCI rats were stratified to undergo no-ABPT, two 20-min bouts/day of quadrupedal bodyweight-supported treadmill training (qBWSTT), or hindlimb passive isokinetic bicycle (cycle) training, 5 days/wk for another 3 wk. We assessed locomotor recovery and plantar flexor muscle mass, tracked cancellous and cortical bone microstructure at the distal femoral and proximal tibial epiphyses using in vivo microcomputed tomography (microCT), and evaluated bone turnover at the tibial epiphysis with histomorphometry. All SCI animals displayed persistent hindlimb paralysis and pervasive muscle atrophy. Over the initial 2 wk, which included 1 wk of no exercise and 1 wk of ABPT acclimation, a similar magnitude of bone loss developed in all SCI groups. Thereafter, cancellous bone loss and cortical bone decrements increased in the SCI no-ABPT group. qBWSTT attenuated this trabecular bone loss but did not prevent the ongoing cortical bone deficits. In comparison, twice-daily cycle training increased the number and activity of osteoblasts versus other SCI groups and restored all bone microstructural parameters to SHAM levels at both epiphyseal sites. These data indicate that a novel passive isokinetic cycle training regimen reversed cancellous and cortical bone deterioration at key epiphyseal sites after experimental SCI via osteoblast-mediated bone anabolic mechanisms, independent of locomotor recovery or increased muscle mass.NEW & NOTEWORTHY This study was the first to assess how quadrupedal bodyweight-supported treadmill training or passive isokinetic bicycle (cycle) training impacts bone recovery at the distal femoral and proximal tibial epiphyses in a rat model of severe contusion spinal cord injury. Our results demonstrate that passive isokinetic cycle training completely restored cancellous and cortical bone microstructural parameters at these sites via osteoblast-mediated bone anabolic actions, independent of locomotor recovery or increased plantar flexor muscle mass.