A high correlation has been documented between the left and right femoral bone mineral densities in the normal population. This suggests that dual femur measurements are not justified in clinical practice. This study evaluated whether this premise holds for subjects who have lost bone mass and have sustained fractures with minimal trauma. Seventy-eight women aged 31-83 years (mean=66 years) with previous low-impact fractures had both proximal femora measured using dual-energy X-ray absorptiometry. There were significant correlations between values in the left and right total hip (TH) (r=0.95; p<0.05) and in the left and right femoral neck (FN) (r=0.90; p<0.05). The mean differences between the left and right TH and FN densities were not significant. However, the range of the limits of agreement for the TH (-0.074 to 0.086 g/cm2) and FN (-0.115 to 0.105 g/cm2) were greater than the 95% confidence interval for true change for the TH (0.05 g/cm2) and FN (0.07 g/cm2). Any longitudinal BMD assessment therefore needs to measure the same proximal femur to get a reliable comparison. A one-tailed analysis showed that for the TH, 7.5% of subjects had a T-score discordance greater than or equal to 0.5 and 0.5% had a T-score discordance greater than or equal to 1. For the FN, 9% had a T-score discordance greater than or equal to 0.5 and 2.5% had a T-score discordance greater than or equal to 1. The use of dual femur measurements increases the diagnostic yield by about 10% in subjects with prior minimal trauma fractures.