The purpose of this study was to compare magnetic resonance imaging (MRI) features and proton MR spectroscopy (1H-MRS) patterns of multiple sclerosis (MS) plaques in order to define the metabolic substrate in different lesion subtypes. Combined MRI and single-voxel 1H-MRS investigation was performed in 54 MS patients (47 relapsing remitting (RR) and seven secondary progressive (SP)). Sixty-seven MS lesions were selected. Thirty-seven lesions were Gadolinium (Gd) enhancing (nine isointense and 28 hypointense on pre-contrast T(1)-weighted scans) and 30 Gd unenhancing (six isointense and 24 hypointense on pre- and post-contrast T(1)-weighted scans). Choline (Cho), creatine (Cr), N-acetyl aspartate (NAA) and lactate were evaluated in 1H spectra acquired from MS plaques and from normal white matter (NWM) of 22 neurological controls. MS lesions of RR patients were characterized by a significant increase of Cho/Cr and decrease of NAA/Cr and NAA/Cho ratios. No significant metabolite changes were found in lesions of SP patients. Gd enhancing plaques showed lactate signal with higher frequency (37.8%) than Gd unenhancing plaques (16.7%) (p=0.04). A significant increase of Cho/Cr was found in Gd enhancing lesions when compared to controls (p<0.01), and to Gd unenhancing lesions (p<0.05). In particular, there was evidence of a significant increase of Cho/Cr in pre-contrast T(1) hypointense Gd enhancing lesions (p<0.01 vs. controls). The Gd unenhancing lesions (p<0.01), in particular the T(1) hypointense group (p<0.05), showed a significant decrease of NAA/Cr only when compared to controls. These data confirm that in vivo MRS indicates key pathological features of MS plaques. The increased Cho/Cr ratio found in Gd-enhancing plaques, in particular in the T(1) hypointense lesions, may reflect increased membrane cell turnover. The T(1) hypointense Gd unenhancing plaques better reflect axonal damage, as suggested by the decrease of NAA/Cr. Nevertheless, the lack of statistical differences in NAA/Cr between plaque subgroups suggests that axonal impairment might occur even in the early stages.