To compare the accuracy of cone beam CT scanning (NewTom 3G) with intraoral periapical radiography (Dixi2, Planmeca CCD sensor and Insight film) for the detection of periapical bone defects. Ten frozen pig mandibles were used. All soft tissues were removed and the jaws were sagittally sectioned to obtain three blocks from each side of the jaw containing the premolars and the molars with surrounding jaw bone. All teeth with intact roots were then "extracted". First, 15 blocks were used to define defect size and exposure parameters; then, the remaining 45 blocks were divided into three equal groups. In one group, cylindrical defects of 1 x 1 mm were prepared beyond the apices of the extraction sockets, in another group defects of 2 x 2 mm were similarly prepared, while no defects were prepared in the last group. The teeth were replaced into their sockets and digital and conventional radiographs of all blocks were taken under reproducible conditions. In addition, all blocks were CT scanned with the same volumetric data and then reconstructed to provide sagittal and coronal 2-D sections. Masked evaluation of the images (defect present vs no defect) was performed by four calibrated examiners. Statistical analysis was performed with ANOVA and the significance level was set to P < 0.05. NewTom 3G was statistically significantly better in terms of sensitivity (54%), positive (82.6%) and negative (44.5%) predictive values, and diagnostic accuracy (61%) when compared with digital radiographs (23%, 60%, 31%, 39%), and with conventional ones (28%, 70%, 35%, 44%)-except in the positive predictive value. Specificity was similar for all three methods. No difference was observed between the two periapical (digital vs conventional) radiographs. NewTom 3G may be useful in cases of immediate implants intended to replace teeth with suspicion for possible existing endodontic pathology, or in candidate implant sites neighboring such teeth.