Subperiosteal implants are currently fabricated by using the classic two-stage direct bone impression technique or by the use of the one-stage computer tomography/computer-assisted designcomputer-assisted manufacture (CT/ CAD-CAM) method. This study compares the accuracy of the two techniques by using cadaver maxillae and mandibles as the models for fabrication of casts. Seven cadaver jaw specimens were collected and subjected to direct bone impressions and to CT scans. Those derived from the direct bone impressions were poured in die stone, while the CT scans were sent for fabrication of CAD-CAMgenerated casts. On each of the 14 models so produced, a cast grid was fabricated that was designed as a measuring device. The preciseness of fit of each grid was subjected to analyses that presented levels of accuracy. Statistical evaluation of these levels, reduced to numerical indices, revealed that the direct bone techniques resulted in acceptable castings in seven of seven cases, whereas the CAD-CAM method yielded adequate castings in five of seven cases.