Purpose: To develop a method for capturing the three-dimensional (3D) shapes of dental arches in a position relative to that of maximum intercuspation and to evaluate its dimensional and occlusal accuracy. Materials and Methods: The conventional custom-tray impression technique was used to capture detailed and dimensionally accurate impressions of individual teeth, and a modified bite-registration technique was used to register the relative positions of the maxillary and mandibular teeth in maximum intercuspation. It was hypothesized that this procedure may help to eliminate the occlusal inaccuracy caused by mouth opening-induced mandibular flexion. Two types of rigid frames (buccal and palatal) were used to prevent deformation of the interocclusal record. Their effects were tested on an articulated full-arch master cast and compared in terms of dimensional accuracy. In addition, the procedure was applied to a healthy volunteer to visually evaluate occlusal accuracy based on the form and distribution of the occlusal contacts. Results: The mean decrements of the dental arch width were 0.037 ± 0.017 mm and 0.269 ± 0.114 mm when using the palatal and buccal frames, respectively. The dimensional accuracy of the palatal frame was comparable to that of the custom-tray impression technique. The form and distribution of the occlusal contacts between the 3D dental arches were similar to those observed in the transilluminated image of the interocclusal record, indicating the occlusal accuracy of this method. Conclusion: The dimensional and occlusal accuracy of the method proposed here is suitable for clinical application when used in combination with the palatal frame.