In vivo 1H MR spectra of the prefrontal cortex acquired with the stimulated echo acquisition mode (STEAM) TE = 20 ms sequence were quantified to determine relative levels of cerebral metabolites. A priori knowledge of spectra from individual metabolites in aqueous solution was incorporated into a frequency domain quantification technique. The accuracy and precision of modeling these metabolites were investigated with simulated spectra of varying signal-to-noise ratios (SNRs) and relative metabolite levels. The efficacy of modeling in vivo data was tested by quantifying 10 repeated measures of two consecutively acquired in vivo spectra (an 8-cm3 volume of interest (VOI) and a 4-cm3 VOI positioned within the 8-cm3 VOI) on the same normal subject. The differences in levels of glutamate (Glu), phosphocreatine plus creatine (PCr+Cr) and choline-containing compounds (Cho(t)) between spectra from the 8- and 4-cm3 VOIs corresponded with the expected differences observed in the proportions of gray matter within the VOIs (estimated from 1H images). Correcting for the T1 and T2 relaxation, the estimated concentrations of N-acetylaspartate, PCr+Cr, Cho(t), Glu, and glutamine were consistent with previous in vivo and in vitro reports.