Bone infection results from bacterial invasion, which is extremely difficult to treat in clinical, orthopedic, and traumatic surgery. The bone infection may result in sustained inflammation, osteomyelitis, and eventual bone non-union. Establishment of a feasible, reproducible animal model is important to bone infection research and antibiotic treatment. As an in vivo model, the rabbit model is widely used in bone infection research. However, previous studies on rabbit bone infection models showed that the infection status was inconsistent, as the amount of bacteria was variable. This study presents an improved surgical method for inducing bone infection on a rabbit, by blocking the bacteria in the bone marrow. Then, multi-level evaluations can be carried out to verify the modelling method. In general, debriding necrotic tissue and implantation of vancomycin-loaded calcium sulphate (VCS) are predominant in antibiotic treatment. Although calcium sulphate in VCS benefits osteocyte crawling and new bone growth, massive bone defects occur after debriding. Autogenous bone (AB) is an appealing strategy to overcome bone defects for the treatment of massive bone defects after debriding necrotic bone. In this study, we used the tail bone as an autogenous bone implanted in the bone defect. Bone repair was measured using micro-computed-tomography (micro-CT) and histological analysis after animal sacrifice. As a result, in the VCS group, bone non-union was consistently obtained. In contrast, the bone defect areas in the VCS-AB group were decreased significantly. The present modeling method described a reproducible, feasible, stable method to prepare a bone infection model. The VCS-AB treatment resulted in lower bone non-union rates after antibiotic treatment. The improved bone infection model and the combination treatment of VCS and autogenous bone could be helpful in studying the underlying mechanisms in bone infection and bone regeneration pertinent to traumatology orthopedic applications.