Quantitative metabolomics of microbial cultures requires well-designed sampling and quenching procedures. We successfully developed and applied a differential method to obtain a reliable set of metabolome data for Escherichia coli K12 MG1655 grown in steady-state, aerobic, glucose-limited chemostat cultures. From a rigorous analysis of the commonly applied quenching procedure based on cold aqueous methanol, it was concluded that it was not applicable because of release of a major part of the metabolites from the cells. No positive effect of buffering or increasing the ionic strength of the quenching solution was observed. Application of a differential method in principle requires metabolite measurements in total broth and filtrate for each measurement. Different methods for sampling of culture filtrate were examined, and it was found that direct filtration without cooling of the sample was the most appropriate. Analysis of culture filtrates revealed that most of the central metabolites and amino acids were present in significant amounts outside the cells. Because the turnover time of the pools of extracellular metabolites is much larger than that of the intracellular pools, the differential method should also be applicable to short-term pulse response experiments without requiring measurement of metabolites in the supernatant during the dynamic period.