We assessed the clinical usefulness of bone density measurements at the os calcis as a screening tool to identify patients with low bone density at the lumbar spine and femoral neck. Bone mineral density (BMD) was recorded in 443 women (mean age 60 years) referred to a bone densitometry service. Measurements were made at the lumbar spine and femoral neck using a Lunar DPXL and at the right os calcis using a Peripheral Instantaneous X-ray Imaging (PIXI) dual-energy X-ray absorptiometry system. Average T-scores derived using the manufacturer's data were: 1.59 for the lumbar spine, -1.41 for the femoral neck and -0.87 for the os calcis. The prevalence of osteoporosis using WHO criteria (T-scores of -2.5 or less) was 36% for the lumbar spine or femoral neck but only 9.7% for the os calcis. BMD of the os calcis correlated with that at the lumbar spine (r = 0.69, p < 0.001) and femoral neck (r = 0.67, p < 0.001). The area under the receiver operator characteristics curve was 0.836 (standard error 0.020) for the os calcis related to osteoporosis at the lumbar spine or femoral neck. Optimal accuracy was obtained at a T-score of < or = -1.3 (BMD 0.39 g/cm2) when the sensitivity was 69.6% (95% confidence interval 65.3, 73.9%) and specificity 82.6% (95% confidence interval 79.1, 86.1%). However, the probability of diagnosing low bone density from a given BMD at the os calcis varied by age and site scanned. Accordingly, for informing management strategies, the choice of a single cutoff BMD at the os calcis may not be appropriate and several thresholds may be adopted based on age, the site of interest (lumbar spine or femoral neck) and consideration of associated clinical features. Thus, the use of heel bone density scanners could reduce the number of axial bone density measurements required. The advantages of portability, low cost and shorter scan times should reduce the cost of detection and provide a greater opportunity for identification of women at risk of fracture.