The treatment of maxillofacial tumors requires coordinated resection and reconstruction. Conventional template-guided open osteotomy techniques may increase the risk of tissue injury. Additionally, the accuracy of titanium plate-based reconstruction is limited by the lack of standardized positional criteria, thereby potentially affecting alignment consistency. To address these challenges, we propose a maxillofacial tumor treatment system termed RAMRS, comprising 2 modules: robot-assisted osteotomy and augmented reality-guided reconstruction (ARR). Within therobot-assisted osteotomy module, a hand–eye calibration framework is used to support robotic arm positioning, while preoperative CT-to-patient spatial alignment is achieved through an optical probe-mediated registration method, supporting robot-assisted osteotomy according to the preoperative plan. The ARR module incorporates a rotating-caliper-based calibration compensation framework to reduce marker registration errors caused by manual instability or fiducial marker deformation. Subsequently, quick-response markers are used to achieve spatial alignment between the preoperative 3-dimensional model and the intraoperative defective mandible, and the virtual mandibular model is projected onto a display visible to the surgeon, thereby providing intuitive intraoperative guidance. Validation experiments were conducted on cadaveric and ex vivo specimens. The results demonstrated favorable accuracy in osteotomy and reconstruction alignment in the evaluated experiments, while the ARR module achieved low 2-dimensional fusion error for augmented reality visualization. These findings support the feasibility of the proposed workflow in preclinical settings.