An integrated sensing system is presented for the first time, where a metal oxide semiconductor sensor-based electronic olfactory system (MOS array), employed for pathogen bacteria identification based on their volatile organic compound (VOC) characterisation, is assisted by a preliminary separative technique based on gravitational field-flow fractionation (GrFFF). In the integrated system, a preliminary step using GrFFF fractionation of a complex sample provided bacteria-enriched fractions readily available for subsequent MOS array analysis. The MOS array signals were then analysed employing a chemometric approach using principal components analysis (PCA) for a first-data exploration, followed by linear discriminant analysis (LDA) as a classification tool, using the PCA scores as input variables. The ability of the GrFFF-MOS system to distinguish between viable and non-viable cells of the same strain was demonstrated for the first time, yielding 100 % ability of correct prediction. The integrated system was also applied as a proof of concept for multianalyte purposes, for the detection of two bacterial strains (Escherichia coli O157:H7 and Yersinia enterocolitica) simultaneously present in artificially contaminated milk samples, obtaining a 100 % ability of correct prediction. Acquired results show that GrFFF band slicing before MOS array analysis can significantly increase reliability and reproducibility of pathogen bacteria identification based on their VOC production, simplifying the analytical procedure and largely eliminating sample matrix effects. The developed GrFFF-MOS integrated system can be considered a simple straightforward approach for pathogen bacteria identification directly from their food matrix. [Figure not available: see fulltext.]