Objective: Manual measurement of mandibular cortical width has been shown to be significantly correlated with bone mineral density (BMD) at the hip,spine and forearm.These latter sites are fracture-prone in post-menopausal osteoporosis. We have developed computer software technology, based on active shape modelling, that will automatically detect the mandibular cortex on panoramic radiographs and then measure its width. We wanted to compare manual measurement of mandibular cortical width with semi-automated constrained fit and automated unconstrained fit methods. Material and Methods:The diagnostic ability of these different methods at detecting osteoporosis of the femoral neck was compared using dual energy X-ray absorptiometry as the “gold standard “at these sites. 671 subjects were recruited and their BMD measured at the femoral neck.143 (21.4%) subjects were osteoporotic, classified as a standardized t-score of <-2.5 at any site. The ability of one observer’s manual measurement of mandibular cortical width to act as a diagnostic test for detecting osteoporosis at the femoral neck was assessed using ROC analysis. The AUC under the ROC curve was = 0.748, (se = 0.027; 95% CI = 0.713 to 0.780). Using a previous training set of 132 DPR images, a point distribution model (PDM) was created of the inferior mandibular cortex. Two experts performed this task independently using a graphical user interface, outlining the inferior mandibular cortex by placing equally spaced points on the computer images between the mental foramen and antegonial region. In the constrained fit, four manually defined reference points were placed on the inferior mandibular surface of the images at the left and right antegonial and mental foramen regions. These points served as starting points for the active shape model and the point distribution model was stretched and positioned to match these points. Results:As a diagnostic test in detecting osteoporosis at the hip, the semi- automated constrained fit produced an AUC of 0.755 (se= 0.026; 95% CI = 0.721 – 0.788). In detecting osteoporosis at the hip, the automated unconstrained fit produced an AUC of 0.739 (se = 0.028; 95% CI = 0.704 to 0.772). There were no significant differences in AUC between the manual, and either the constrained or unconstrained fits (p>0.05).