Adenovirus vector system is expected to be useful for direct gene therapy for joint disease. This study first sought to confirm that foreign genes can be transferred to articular chondrocytes in primary culture. Next, recombinant adenovirus vectors harbouring beta-galactosidase gene (LacZ) was injected directly into the temporomandibular joints of Hartley guinea-pigs to clarify the in vivo transfer availability of the adenovirus vectors. Specifically, recombinant adenovirus harbouring LacZ gene (AxlCALacZ) was injected into the upper joint cavities of both mandibular joints of four male 6-week-old Hartley guinea-pigs. Either the same amount of recombinant adenovirus without LacZ gene (Axlw) suspension (placebo) or the same amount of phosphate-buffered saline solution (control) were injected into the upper joint cavities of both joints of another four male guinea-pigs. At 1, 2, 3 and 4 weeks after injection, the joints were dissected and the expression of delivered LacZ was examined by 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside (X-gal) staining and reverse transcriptase-polymerase chain reaction (RT-PCR). To investigate the expression of transferred gene in other organs, total RNA was extracted from liver, kidney, heart and brain and the expression of LacZ mRNA and 18 S ribosomal RNA were analysed by RT-PCR. Clear expression of LacZ was observed in the articular surfaces of the temporal tubercle, articular disc and synovium of the temporomandibular joints even 4 weeks after injection in the AxlCALacZ-injected group, while no expression was detected in placebo and control groups. Histological examination confirmed that LacZ activity was clearly detected in a few cell layers of the articular surface tissues, which is much more efficient than in a previously study of the knee joint. In the other organs, expression of the delivered transgene was not observed. Based on these findings, direct gene delivery into the articular surface of the temporomandibular joint using the adenovirus vector is feasible as an effective in vivo method.