A rigorous quality control platform was developed to monitor short- and long-term retention time reproducibility, instrument response and mass accuracy for untargeted differential analysis of beer. The optimized workflow utilized a QC beer and external calibration standards to yield LC retention time variation <3 s, average interday response RSDs <15 % and average mass errors <10 ppm on a quadrupole time of flight mass spectrometer. QC statistics guided data processing, including molecule feature (MF) extraction, MF alignment across samples and recursion, prior to differential analysis of single-hop India Pale Ales produced in two different production years (2010 vs. 2011). Differential analysis and significance testing revealed that the year of production (vs. hop used) accounted for the greatest compositional difference between samples. Negative and positive ion ESI data collectively revealed a minimum 3× fold change (p < 0.05) for ~4000 MFs. Since identification of ~4000 MFs was untenable, focus turned to the identification of ~150 high-abundance MFs. Targeted MS/MS, Metlin database searching and the use of ChemSpider with automated MS/MS spectral prediction and correlation, identified candidate compounds that were confirmed by authentic standards. Most interestingly, the 2010 beers, which were stored longer at room temperature prior to analysis, provided additional evidence for the use of purines, specifically 5-methylthioadenosine with methionine, as nonvolatile molecular indicators of beer oxidation during storage. While this untargeted metabolomics platform was developed with beer in mind, the workflow described here could be applied to any food metabolomics application.