An assay method for identification of metabolites from in vitro microsomal incubations was developed for use in the early stage of drug discovery. We have developed a practical approach which involves integrated sample generation, sample preparation, bioanalysis, and data handling to maximize sample throughput and speed up the process for identification of metabolites. The assay system consisted of a robotic liquid handler (Genesis workstation) to generate and process samples, PALLAS MetabolExpert software to predict possible metabolites, exact mass measurement via a tandem quadrupole time-of-flight mass spectrometer (QTOF-MS) coupled with liquid chromatography to analyze samples, MetaboLynx software to find potential metabolites and Advanced Chemistry Development / MS (ACD / MS) software to provide guidance to the most likely hypothetical metabolite chemical structures. For purposes of evaluating this new method, dextromethorphan, alprenolol, and propranolol were incubated separately for up to 60 minutes with rat and human hepatic microsomes. The incubation and sample preparation were carried out in 96-well plates using the Genesis workstation. The bioanalysis was performed by LC-MS / MS using QTOF with MetaboLynx software to find metabolites. Metabolic products formed in vitro by rat and human microsomes were separated using an analytical column C18 with gradient elution at flow rate of 250 μl / min. The internal mass calibration was performed by continuous postcolumn infusion of Haloperidol. The mass spectra from incubations containing NADPH were compared to those without NADPH (control) using the MetaboLynx software to find potential metabolites. Finally, the MS / MS spectra were processed by the ACD / MS software to predict the chemical structure. MetaboLynx software successfully identified metabolites for each of the drugs studied by automatically discerning expected metabolites. Exact differences in masses between each metabolite and parent drug were measured from five replicate sample injections. All measured values are accurate to less than 0.001Da or 3.8 ppm with the standard deviation within 0.0015 Da, which allowed good prediction / confirmation of empirical formulae. Hypothetical chemical structures were achieved by the ACD / MS software and provided a useful tool to assist in prediction of the metabolic pathways of the drugs. The metabolites identified were in good agreement with previously published results for all three compounds. This new method will greatly enhance throughput, which in turn will facilitate our ability to rapidly provide this guidance to the synthetic chemist.