The full range of fracture risk determinants arise from each hierarchical level comprising the organization of bone. Raman spectroscopy is one tool capable of characterizing the collagen and mineral phases at a near submicron-length scale, but the ability of Raman spectra to distinguish compositional differences of bone is not well defined. Therefore, we analyzed multiple Raman peak intensities and peak ratios to characterize their ability to distinguish between the typically less mineralized osteonal tissue and the more mineralized interstitial tissue in intracortical human bone. To further assess origins of variance, we collected Raman spectra from embedded specimens and for two orientations of cut. Per specimen, Raman peak intensities or ratios were averaged among multiple sites within five osteons and five neighboring interstitial tissue. The peak ratios of ν(1) phosphate (PO(4)) to proline or amide III detected the highest increases of 15.4 or 12.5%, respectively, in composition from osteonal to interstitial tissue. The coefficient of variance was less than 5% for each as opposed to a value of ~8% for the traditional ν(1)PO(4)/amide I, a peak ratio that varied the most between transverse and longitudinal cuts for each tissue type. Although embedding affected Raman peaks, it did not obscure differences in most peak ratios related to mineralization between the two tissue types. In studies with limited sample size but sufficient number of Raman spectra per specimen for spatial averaging, ν(1)PO(4)/amide III or ν(1)PO(4)/proline is the Raman property that is most likely to detect a compositional difference between experimental groups.