OBJECTIVE To compare proximal caries detection using intraoral bitewing, extraoral bitewing and panoramic radiography. METHODS 80 extracted human premolar and molar teeth with and without proximal caries were used. Intraoral radiographs were taken with Kodak Insight film (Eastman Kodak Co., Rochester, NY) using the bitewing technique. Extraoral bitewing and panoramic images were obtained using a Planmeca Promax Digital Panoramic X-ray unit (Planmeca Inc., Helsinki, Finland). Images were evaluated by three observers twice. In total, 160 proximal surfaces were assessed. Intra- and interobserver kappa coefficients were calculated. Scores obtained from the three techniques were compared with the histological gold standard using receiver operating characteristic analysis. Az values for each image type, observer and reading were compared using z-tests, with a significance level of α = 0.05. RESULTS Kappa coefficients ranged from 0.883 to 0.963 for the intraoral bitewing, from 0.715 to 0.893 for the extraoral bitewing, and from 0.659 to 0.884 for the panoramic radiography. Interobserver agreements for the first and second readings for the intraoral bitewing images were between 0.717 and 0.780, the extraoral bitewing readings were between 0.569 and 0.707, and the panoramic images were between 0.477 and 0.740. The Az values for both readings of all three observers were highest for the intraoral bitewing. Az values for the extraoral bitewing images were higher than those of the panoramic images without statistical significance (p > 0.05). CONCLUSIONS Intraoral bitewing radiography was superior to extraoral bitewing and panoramic radiography in diagnosing proximal caries of premolar and molar teeth ex vivo. Similar intra- and interobserver coefficients were calculated for extraoral bitewing and panoramic radiography.