An extensive series of quantitative simulation studies replicating ideal dual-energy X-ray absorptiometric (DXA) bone mineral density (BMD) measurements of typical and realistic in vivo lumbar vertebral and proximal femoral sites has been carried out to quantitatively assess the extent of inherent systematic inaccuracies in such measurements. The results for these bone sites indicate that BMD inaccuracies as high as 20% or more can be anticipated clinically, particularly in cases of osteopenic, osteoporotic, and elderly patients. It is found that the most important soft-tissue anthropometric determinants of the extent of bone site-specific systematic in vivo BMD inaccuracies reflected in DXA measurements are the ratio of the areal density of extraosseous fat to that of lean muscle tissue immediately surrounding the interrogated bone site and the specific yellow/red marrow mix within the scanned bone. As such, the present findings focus directly on the question of whether or not the strong, positive correlations and associations between soft tissue compositional parameters and DXA-measured in vivo BMD determined in a large number of previous clinical investigations are, in toto or in part, biologically causal. The present results are seen to be quantitatively and qualitatively in conformity with the many clinical studies that have found marked general decreases (increases) in measured BMD as body weight and/or body fat mass decreases (increases). It is concluded that the clinically observed correlations between DXA-measured BMD and these anthropometric parameters are artefacts of the systematic errors (inaccuracies) inherent in planar DXA methodology and are unlikely to be of biological genesis.