In this study, the in vivo reproducibility of stochastic predictors and BMD values from Dual- Energy X-ray Absorptiometry (DXA) were assessed at two different anatomic sites:​ hip and spine. Measurement of bone mineral density (BMD) derived from DXA has been widely used as a diagnostic tool for osteoporosis. BMD has been used as a representative value for bone strength and fracture risk, but only represents a part of both. In order to better understand fracture risk, the use of grayscale images from DXA scans has been implemented in enhancing the prediction of bone strength and fracture risk. The use of stochastic predictors derived from DXA scans is a recent method of evaluating the bone quality and fracture risk. The goal of this study was to examine the precision of stochastic predictors and BMD values from DXA scans. The study involved 15 postmenopausal women over the age of 55 with no history of osteoporotic fractures. Three sets of in vivo DXA scans were taken at 2 different scan sites (hip and spine) in the posterior-anterior projection. An International Society for Clinical Densitometry certified technician performed the scans on a Hologic densitometer (QDR Discovery W, Bedford, MA). Lumbar spine and hip values of BMD were assessed for precision from the DXA program. Using the same data from the DXA scans, an experimental variogram was used to derive stochastic predictors. The precision values, expressed as RMS-%CV, for the BMD measurement were computed. The RMS-%CV values of stochastic predictors for the hip and lumbar spine was also used to determine reproducibility of measurement. The study found that although the lumbar vertebrae and the total hip RMS-%CV were within the recommended values as outlined by the ISCD, the stochastic predictors did not coincide in precision. The stochastic predictors are a relatively new imaging process that has no other study to compare to. The significance of the RMS-%CV in the stochastic predictors is limited by the amount of data available and the nature of inhomogeneity, represented by stochastic predictors.