METHODS Prospective study. OBJECTIVE To assess the proton MR spectroscopy (1H MRS) of vertebral bone marrow and correlate the lipid water ratio (LWR) and spectral line width (LW) with bone mineral density (BMD) in female subjects. BACKGROUND The mechanism of bone marrow fat accumulation and bone mineral content is poorly understood. Proton MR spectroscopy was used to demonstrate the lipid and water spectra in the bone marrow. We try to assess the possible interaction between the bone marrow lipid content, aging, and BMD. METHODS Proton MRS and BMD of the lumbar spine were performed in 52 female subjects (mean age, 58 years; SD, 10 years). They were 13 premenopausal and 39 postmenopausal women. The BMD (g/cm2) was measured using dual energy radiograph absorptiometry at the lumbar spine. Single voxel 1H MRS was measured at L3 vertebral body by stimulated echo-acquisition mode (STEAM) sequence and demonstrated two major peaks (lipid and water). Comparisons of the differences between the two subgroups were made. Pearson's correlation was also calculated to explore the association of 1H MRS measurements with age and BMD. Partial correlation was further conducted when controlling the variable such as age or BMD. RESULTS BMD and LWR had statistically significant difference between the pre- and postmenopausal subgroups (P < 0.001), while lipid LW had a borderline difference and water LW had no difference. LWR was positively correlated with age (r = 0.52 and P < 0.0001) and negatively correlated with BMD (r = -0.40 and P = 0.003) for all the subjects. Lipid LW was negatively correlated with age (r = -0.32 and P = 0.0197) and positively correlated with BMD (r = 0.67 and P < 0.0001). When controlling for BMD effect, only LWR is statistically correlated with age (partial r = 0.39, P = 0.0045), while only the lipid LW is statistically correlated with BMD when controlling for age (partial r = 0.63, P < 0.0001). None of the correlations between water LW and age or BMD was significant. In the subgroups, only the lipid LW is significantly correlated with BMD (r = 0.78, P = 0.0016 in premenopausal women; r = 0.62, P < 0.0001 in postmenopausal women). CONCLUSIONS The LWR had a positive correlation with the age, while the lipid LW had a positive correlation with BMD, even after controlling the age factor. The bone marrow lipid water content and metabolism acted as important roles in the internal environment of bone and influenced bone mineralization.