The aim of this study was to evaluate retrospectively the accuracy of a protocol for completely in-house, computer-assisted, orthognathic surgery, generating resin printed intermediate surgical guides. A retrospective, observational study was made on a cohort of 15 patients treated consecutively from September 2017 to May 2019, who underwent bimaxillary orthognathic surgery planned with the same 3-dimensional program and whose surgical intermediate splints were obtained with the same all-in-house protocol. Virtual planned surgical movements were compared with the real surgical outcome. The differences were not significant for eight of the 12 variables considered. The p values, calculated with the Wilcoxon signed rank test, were evenly distributed and ranged from p=0.001 to p=0.820. Significant differences were reported in four measurements: angle between sella-nasion plane and a line connecting the incisal edge and the apex of the root of the most prominent incisor (U1-SN) (p=0.001), angle between Frankfort plane and a line connecting the incisal edge and the root apex of the most prominent upper incisor (p=0.008), dental midline discrepancies (p=0.006), and occlusal plane tilt (U1-FH) (p=0.001), basically due to intraoperative settings. The 3-dimensional resin printed surgical guides were shown to be a reliable alternative to the commercial ones and showed high rate of accuracy for most of the variables assessed. Four out of 12 of these showed significant errors, but two of them were only minimal discrepancies with no clinical implications.