We have developed a novel metabolic modeling methodology that traces the flow of functional moieties (chemical substructures) through metabolic pathways via the deconvolution of mass isotopologue data of specific metabolites. We have implemented a general simulated annealing/genetic algorithm for parameter optimization called Genetic Algorithm for Isotopologues in Metabolic Systems (GAIMS), with a model selection method developed from the Akaike information criterion. GAIMS is tailored for analysis of ultrahigh resolution, high mass-accuracy isotopologue data from Fourier transform-ion cyclotron resonance mass spectrometry (FT-ICR-MS) for interpretation of non-steady state stable isotope-resolved metabolomics (SIRM) experiments. We applied GAIMS to a time-course of uridine diphospho-N-acetylglucosamine (UDP-GlcNAc) and uridine diphospho-N-acetylgalactosamine (UDP-GalNAc) isotopologue data obtained from LNCaP-LN3 prostate cancer cells grown in [U-13C]-glucose. The best metabolic model was identified, which revealed the relative contribution of specific metabolic pathways to 13C incorporation from glucose into individual functional moieties of UDP-GlcNAc and UDP-GalNAc. Furthermore, this analysis allows direct comparison of MS isotopologue data with NMR positional isotopomer data for independent experimental cross-verification.