The validity of the bone mineral density (BMD) measurement depends on its accuracy as a predictor of the breaking strength of bone. As the breaking strength is proportional to the square of the apparent density, a small error in the calculation of BMD may result in a larger error in the predicted bone strength. The aims of this study were (i) to determine whether inaccuracies in the measurement of the dimensions, projected area, and volume of the vertebral body (used to derive the areal and volumetric BMD) result in errors in the predicted breaking strength and (ii) to compare the accuracy, sensitivity, and specificity of bone mineral content (BMC), areal BMD, volumetric BMD, and volumetric bone mineral apparent density (BMAD) as surrogates of bone strength. We measured the BMC (by densitometry), dimensions and volume (using calipers, densitometry, the Carter et al. and Peel and Eastell methods), and breaking strength (using the Instron 1114 apparatus, Newtons, N) of 22 vertebral body specimens. All methods resulted in errors in height, width, and depth between -11.3 +/- 1.0 and 30.4 +/- 1.8% relative to the "gold" standard caliper method. The vertebral body volume (of 38.0 +/- 1.2 cm3) measured by submersion was used as the gold standard to derive the volumetric BMD gold standard (of 0.162 +/- 0.01 g/cm3). All methods, except the Peel and Eastell method, resulted in errors ranging between -10.7 +/- 1.5 and 56.9 +/- 3.4% in vertebral body volume and -35.6 +/- 1.5 to 12.6 +/- 1.8% in volumetric BMD (all p < 0.0005). The same absolute value for volumetric BMD predicted a breaking strength that differed according to the method used to derive BMD. For example, a volumetric BMD of 0.162 g/cm3 predicted a breaking strength of 6208 N (submersion method), 5473 N (caliper method), 6095 N (Peel and Eastell method), 7697 N (DXA method), and 9470 N (Carter et al. method). The mean volumetric BMD derived by each method differed (0.181, 0.165, 0.133, and 0.104 g/cm3, respectively). However, all were accurate; each predicted a similar breaking strength (6177, 6217, 6209, and 6221 N respectively). Likewise, breaking strengths predicted by the mean BMC, areal BMD by calipers, and areal BMD by dual-energy X-ray absorptiometry (DXA) were 6267, 6214, and 6244 N, respectively. The methods were equally sensitive; a 1 standard deviation (SD) decrease in volumetric BMD resulted in a similar decrease in the breaking strength of 1818 (caliper), 2080 (Peel and Eastell), 2001 (DXA), and 1625 N (BMAD by Carter et al). A 1 SD decrease in BMC, areal BMD (using calipers) and areal BMD (using DXA) predicted a decrease in the breaking strength of 2019, 1738, and 1825 N, respectively. All methods were equally specific; the variance in bone strength explained by bone mass did not differ for volumetric BMD (38-61% depending on the method), BMC (58%), or areal BMD (48%). In conclusion, despite errors in the measurement of the dimensions of the vertebral body, bone mass, areal, and volumetric bone density are equally accurate, sensitive, and specific surrogates of the breaking strength of bone in vitro.