Fat-spread products, including margarines, are an emulsion of water derived from vegetable and/or animal fats, which is obtained through a hydrogenation process. In the market there are different types of fat-spreads and their composition depends on the country where they are produced. It is therefore desirable to identify the geographic origin of each product. In this study, a multivariate analysis based on data fusion methodology (low- and mid- level) is carried out using partial least squares-discriminant analysis (PLS-DA) in order to classify fat-spread products according to their geographic origins (Spain, Morocco and France). The instrumental fingerprints acquired using reverse-high performance liquid chromatography coupled with a diode array detector ((RP)HPLC-DAD) and gas chromatography coupled with a mass spectrometry detector (GC-MS) were employed as analytical signals. The performance of the classifications was evaluated using proper classification metrics. Finally, low-level fusion turned out to achieve better results than mid-level fusion, obtaining sensitivity and precision values in external validation equal to 1.00.