This study was designed to confirm relationships between decrease of bone mineral density and increase of marrow fat and to delineate, through MR spectroscopy, vertebral body at high risk for compression fracture onset to justify prophylactic vertebroplasty. We enrolled 127 women: 48 osteoporotic, 36 osteopenic, and 43 normal subjects, who underwent DXA and MR examination of spine. Then, we selected 48 patients with at least two acute osteoporotic vertebral fractures with interposed normal ''sandwich'' vertebrae; all patients underwent MR examination of spine. Significant statistical differences were found among ''Fat Fraction'' (FF) values in normal, osteopenic, and osteoporotic subjects: 59.8 ± 5.1%; 64.8 ± 4.4%; and 67.1 ± 3.3%. A mild, significant, negative correlation was observed between T-score and vertebral fat content (r = -0.585; P = 0.0000). In the second part of the study, 9 new vertebral fractures were observed in 48 patients (19%): 6 were ''sandwich'' vertebrae (12.5%), and 3 were located in distant vertebral body. The mean FF in sandwich fractured vertebrae was 72.75 ± 1.95 compared with the FF of the nonfractured sandwich, and distant control vertebrae were 61.83 ± 3.42 and 61.42 ± 3.64. We found a significant statistical difference between fractured and nonfractured vertebrae (P \ 0.001). The results of this study suggest that MR spectroscopy could be a reliable index to predict the risk of new compression vertebral fracture and could be used for vertebroplasty planning contributing to clarify the possibility to add prophylactic PVP to standard treatment.