Accurate determination of intracellular metabolite levels requires well-validated procedures for sampling and sample treatment. Several methods exist for metabolite extraction, but the literature is contradictory regarding the adequacy and performance of each technique. Using a strictly quantitative approach, we have re-evaluated five methods (hot water, HW; boiling ethanol, BE; chloroform-methanol, CM; freezing-thawing in methanol, FTM; acidic acetonitrile-methanol, AANM) for the extraction of 44 intracellular metabolites (phosphorylated intermediates, amino acids, organic acids, nucleotides) from S. cerevisiae cells. Two culture modes were investigated (batch and chemostat) to check for growth condition dependency, and three targeted platforms were employed (two LC-MS and one GC/MS) to exclude analytical bias. Additionally, for the determination of metabolite recoveries, we applied a novel approach based on addition of (13)C-labeled internal standards at different stages of sample processing. We found that the choice of extraction method can drastically affect measured metabolite levels, to an extent that for some metabolites even the direction of changes between growth conditions can be inverted. The best performances, in terms of efficacy and metabolite recoveries, were achieved with BE and CM, which yielded nearly identical levels for the metabolites analyzed. According to our results, AANM performs poorly in yeast and FTM cannot be considered adequate as an extraction method, as it does not ensure inactivation of enzymatic activity.