High-grade brain tumors are known to have a high rate of glucose (Glc) consumption. Postmortem measurements have suggested that Glc content in experimental brain tumors is relatively low. We used magnetic resonance spectroscopy to investigate this, in vivo, in the brains of seven rats bearing intracerebral C6 gliomas. We combined the high spectral resolution allowed by two-dimensional proton nuclear magnetic resonance with spatial encoding by magnetic field gradient pulses to obtain in vivo maps of Glc, alanine, hypotaurine, aspartate, phosphoethanolamine, Glu/Gln, N-acetylaspartate (NAA), phosphocreatine/creatine (PCr/Cr), cholinecontaining compounds, and lactate (Lac) (some of which are involved in energy metabolism). Compared with normal brain tissue, the main differences found in the gliomas were that Glc, NAA, PCr/Cr, and aspartate concentrations were much lower, whereas concentrations of alanine, hypotaurine, phosphoethanolamine, and Lac were higher, whatever the extent of necrosis. A striking observation is the similarity of the NAA and Glc images: the concentrations of both metabolites are lower in the tumor than they are in the contralateral brain. If Glc was completely absent from the tumor tissue, and if the residual Glc level was due only to a partial volume effect like that for NAA, a neuronal marker, the ratio [Glc] tumor / [Glc]contralateral tissue, should be similar to that found for NAA. The ratio for Glc was 0.48 6 0.22 (6 SD; n 5 6), a ratio similar to that found for PCr/Cr (0.50 6 0.19) but significantly higher than that obtained for NAA (0.29 6 0.07). This observation indicates that a measurable Glc concentration is still present in the tumor tissue. Intense glycolysis in tumor cells may explain the increased production of Lac and alanine and decreased amount of Glc. These nuclear magnetic resonance measurements of metabolite concentrations are complementary to positron emission tomography, which measures Glc consumption.