Whether antibiotics should be included remains greatly debated in Masquelet technique. This study intended to determine the effect of polymethyl methacrylate (PMMA) spacer loaded with different vancomycin concentrations on bone defect repair. Hollow cylindrical spacers consisting of PMMA and varying vancomycin concentrations (0, 1, 2, 4, 6, 8, and 10 g) were prepared. Critical bone defects of rabbits were created at the radial shaft, and spacers were implanted and subsequently intramedullary fixed with retrograde Kirschner’s wires (n = 4 for each vancomycin concentration). After 4 weeks, the induced membranes were opened and cancellous allografts were implanted into the defects. Eight weeks post-operatively, the results of X-ray, histology, and micro-CT revealed that some cortical bone was formed to bridge the gap and the bone marrow cavity was formed over time. Quantitatively, there was more new bone formation in the groups with a relatively lower vancomycin concentration (1–4 g) compared with that in the groups with a higher vancomycin concentration (6–10 g). Our findings suggested that PMMA spacers loaded with relatively lower vancomycin concentrations (1–4 g) did not interfere with new bone formation, whereas spacers loaded with relatively higher vancomycin concentrations (6–10 g) had negative effects on bone formation.