Abstract Introduction The aim of this study was to compare a carbon plate (CP) and a titanium mandibular reconstruction plate (TMRP) in finite element analysis and an animal model. Materials and methods Twelve rabbits were used for this experiment. After a mandible continuity defect was created, either a CP or a TMRP was used for mandibular reconstruction. Postoperatively, daily feed intake amount (DFIA) was measured for 4 weeks. Radiographic images were also acquired to evaluate screw loosening. For the analysis of the stress distribution, a simple continuity defect model was used, and finite element analysis was performed. Results The CP group had 0.80 ± 0.45 lost screws in an animal during the 4 weeks postoperative observation; however, the TMRP group had 1.86 ± 0.69 lost screws ( p  = 0.014). Overall, the 5 out of 5 of rabbits in the CP group and 3 out of 7 in the TMRP group exhibited preoperative levels of DFIA during the 4 week observation ( p  = 0.038). The finite element analysis showed that the stress was more evenly distributed in the CP than in the TMRP model. Conclusions The CP group showed decreased screw loosening and increased recovery of preoperative DFIA compared to the TMRP group in a rabbit model of mandibular continuity defects. Perfect adaptation of CP during the operation could not be achieved in spite of reshaping to the mandibular curvature. This disadvantage of the CP system can be overcome by the prefabricated technique using a prototype model.