We aimed to investigate the reproducibility and accuracy of Radiofrequency Echographic Multi-Spectrometry (REMS) for femoral BMD estimation and the reproducibility and discriminative power of the REMS-derived femoral fragility score. 175 patients with primary and disuse-related osteoporosis were recruited: one femoral Dual-energy X-ray Absorptiometry (DXA) scan and two femoral REMS scans were acquired. No significant test—retest differences were observed for all REMS-derived variables. The diagnostic concordance between DXA and REMS was 63% (Cohen’s kappa = 0.31) in patients with primary osteoporosis and 13% (Cohen’s kappa: −0.04) in patients with disuse-related osteoporosis. No significant difference was observed between REMS and DXA for either femoral neck BMD (mean difference between REMS and DXA: −0.015 g/cm2) or total femur BMD (mean difference: −0.004 g/cm2) in patients with primary osteoporosis. Significant differences between the two techniques were observed in patients with disuse-related osteoporosis (femoral neck BMD difference: 0.136 g/cm2; total femur BMD difference: 0.236 g/cm2). Statistically significant differences in the fragility score were obtained between the fractured and non-fractured patients for both populations. In conclusion, REMS showed excellent test-retest reproducibility, but the diagnostic concordance between DXA and REMS was between minimal and poor. Further studies are required to improve the REMS—derived estimation of femoral BMD.