The purpose of this study was to measure precision of thoracic quantitative computed tomography (QCT) bone mineral density (BMD) and correlation to lumbar spine QCT bone density. We measured the reproducibility of thoracic QCT; two consecutive thoracic QCT scans of the T9, T10, and T11 vertebrae were performed on 95 subjects (49 females, 46 males; mean age, 62.5 years) undergoing coronary scanning. In order to correlate the thoracic to standard lumbar measurement, the subjects also underwent a lumbar QCT scan of the L1, L2, and L3 vertebrae as part of an abdominal aortic scanning study. The variation of thoracic BMD was assessed in different ethnic subgroups. Consecutive thoracic QCT measurements showed good agreement (r=0.98; RMS CV=5.78%). Thoracic bone density was significantly higher than lumbar bone density results (paired t-test, P=0.003), but the two methods correlated well (r=0.86). The regression equation for the relationship between lumbar (X) and thoracic (Y) QCT was Y=0.87X + 22.97. The standard error of estimate was 19.0 mg/cm3. Thoracic QCT from coronary calcium thoracic scans is able to measure BMD with rescan precision and regression errors that are small compared to the biologic variability in the population. Given the relatively small precision error and the reasonable correlation to lumbar BMD, an ancillary assessment of thoracic BMD in a cardiac scan is likely to be a useful assessment of bone mineral status in the general population.