The purpose of this study was to evaluate the accuracy of model surgery prediction after orthognathic surgery and to identify possible errors associated with the prediction process. The study included 46 patients who had undergone orthognathic surgical procedures; 22 in Group A who had had a Le Fort I osteotomy; and 24 in Group B who had had a Le Fort I osteotomy and mandibular setback surgery. The immediate postoperative and preoperative lateral cephalograms were analysed to calculate surgical changes; these were compared with those obtained from model surgery prediction and a statistical analysis was undertaken. The maxilla was more under-advanced and over-impacted anteriorly than predicted by model surgery. The amount of mandibular setback was more than that predicted by model surgery. None of the differences between prediction planning and actual surgical changes was statistically significant at p<0.05. Inaccuracy with the face bow recording, the intermediate wafer, and auto-rotation of the mandible in the supine or anaesthetized patient would appear to be the principal reasons for errors. Inaccuracies are associated with the transfer of prediction planning to model surgery planning and prediction, which should be eliminated to improve the accuracy and predictability of orthognathic surgery.