Fragmentation of simple sugars, using fluidized thermal bed cracking, results in a mixture that contains a variety of polar oxygenated compounds, some yet unknown to the industries. This lack of understanding of the product composition, hinders the downstream processes in conversion of sugars to useful chemicals. The aim of this study was fourfold: to develop a GC–MS identification workflow for carbohydrate fragmentation compounds, to build a library of compounds detected by five GC–MS methods including liquid injection, liquid-liquid extraction, static headspace, solid-phase microextraction and derivatization, to investigate and compare the compound coverage of the five methods, and finally investigate possible reaction pathways for some of the analytes in the mixture. The analysis was done on a mixture of seven representative sugar fragmentation samples. The identification workflow was based on mass spectral match factors (>80 %), retention indices and analytical standards. A total of 389 compounds were detected, out of which, 46 compounds were fully identified (through confirmatory analysis of analytical standards), 87 were tentatively identified (spectral match factors above 80 %), 71 were un-identified with spectral match factors of below 80 % and 185 were discarded (contaminants in blanks etc.). The production pathway for some of the identified compounds in the sugar conversion mixtures were also discussed. In conclusion, solid-phase microextraction detected the highest number of analytes (93), mostly volatile organic compounds and the derivatization GC–MS technique detected the lowest number of analytes (19) but was the only method to detect the largest molecules (C₆-C₁₂) in size. This library of CF products can be used as a suspect-screening database for other biomass fragmentation mixtures.