Using an in vivo adeno-associated virus (AAV)-mediated gene transfer technique, this study evaluated the therapeutic effects of an osteoprotegerin (OPG) transgene against orthopaedic wear debris-induced osteolysis in a long-term murine model. A titanium pin was surgically implanted into proximal tibia of Balb/c mice to mimic a weight-bearing knee arthroplasty, followed by an intra-articular challenge with Ti particles to provoke periprosthetic inflammation and osteolysis. rAAV-hOPG or AAV-LacZ vectors were injected into the prosthetic joint at 3 weeks post-op. The tissues were harvested at 2, 4, 12 and 24 weeks after transduction for histological and molecular analyses. Successful transgene expression at the local site was confirmed by real-time PCR and ELISA. Inflammatory pseudo-membranes were ubiquitously present at the interface between the Ti implant and the surrounding bone in both LacZ and virus-free control groups, while soft tissue was only observed sporadically at the bone–implant interface in the OPG group. A significant reduction in TRAP+ osteoclast numbers was observed in the OPG treatment group. MicroCT assessment indicated a marked reversal in the loss of peri-implant bone mineral density (BMD) in the OPG-transduced group, when compared with the LacZ and virus-free controls. Further, OPG gene modification appeared to reduce local bone collagen loss by a mean of 40%. Real-time PCR examination confirmed that in vivo OPG gene transfer dramatically influenced the periprosthetic tissue gene expression profiles by diminishing the mRNA expression of TNF, IL-1, CPK and RANKL. There were no transgene-associated toxic effects apparent during the experiment, and the PCR detection of transgenes in remote organs such as lungs, kidneys, liver, and muscle of contralateral limb were consistently negative. Overall, rAAV-mediated OPG gene transfer effectively reversed Ti-particle-induced bone resorption in this experimental model. The therapeutic effects may be due to the blockage of local osteoclastogenesis and possibly the down-regulation of RANKL expression.