Because bone mineral density (BMD) measurements at various sites differ in the relative amounts of cortical and trabecular bone that they assess, they also differ in their sensitivity for detecting osteopenia. Lateral spine dual energy x-ray absorptiometry (DXA) allows measurement of BMD of the vertebral bodies, which contain mainly trabecular bone, without contribution from the posterior vertebral elements, which are rich in cortical bone. Thus, we hypothesized that lateral spine DXA would detect osteopenia more frequently than anterior-posterior (AP) spine DXA. To assess the ability of DXA to estimate trabecular bone mass, we compared AP and lateral DXA spine measurements with trabecular bone measurements by quantitative computed tomography (QCT) in 58 patients. We then compared AP vs. lateral spine DXA measurements in 1) 300 women referred for routine bone densitometry, 2) 30 glucocorticoid-treated women, and 3) 44 women with vertebral compression fractures. To compare short term reproducibility, we performed repeat AP and lateral DXA scans in 50 women. The association between QCT and DXA measurements was stronger when DXA measurements were made in the lateral (r = 0.784) or midlateral (r = 0.823) projection than in the AP (r = 0.571) projection. The association of BMD with age was stronger when DXA measurements were made in the lateral (r = 0.536) or midlateral (r = 0.536) projection than in the AP (r = 0.382) projection. The declines in BMD with age for AP, lateral, and midlateral DXA measurements were 0.48%, 0.60%, and 0.88%/yr, respectively. In the women referred for routine densitometry, lateral DXA measurements were significantly (P < 0.05) more abnormal than AP measurements compared with those in young women. This was also true in the women treated with glucocorticoids and women with vertebral compression fractures. Lateral DXA often detected osteopenia in patients whose AP DXA was normal. The 95% confidence limits for changes in BMD attributable to measurement error for AP, lateral, and midlateral DXA were 0.027, 0.038, and 0.057 g/cm2, respectively. These results indicate that lateral DXA measurements identify patients with osteopenia more often than AP DXA measurements, probably because lateral DXA more accurately estimates trabecular bone mass. Short term reproducibility of lateral DXA is nearly as good as that for AP DXA.