The possibility of re-establishing a rigid bone-implant fixation, i.e. osseointegration, after mechanical loosening of titanium implants, was evaluated in the rabbit tibia. Implants were inserted to engage either one (10 mm long, n = 24) or two (16 mm long, n = 24) cortical layers and were allowed to heal for six weeks. A re-entry was then made and 12 test implants in each group subjected to a reverse torque procedure until the integration failed. The remaining nonrotated 24 implants were left as controls. Thereafter all implants were allowed to heal for an additional period of six weeks. At the end of the 12 weeks, the degree of integration was assessed by measuring the removal torque for six test and six control implants in each group of implant lengths. Histomorphometric measurements were also performed on ground sections of the remaining test and control implants. A statistically significant higher removal torque was observed for the monocortical test implants than for the corresponding control implants. No differences were seen for the bicortical implants of either kind, however, no morphological differences could be revealed either, when comparing monocortical and bicortical test and control implants. These results indicate that osseointegrated implants that have been mobilized due to a traumatic disruption of the bone-implant interface, may reintegrate if allowed to heal for an additional period of time.