Many previous reports suggested that conventional antioxidant assays could not forecast antioxidant performance of plant extracts, especially when it came to a real food system such as oil-in-water emulsion. In this study, antioxidant activities of aqueous and ethanolic plant extracts are investigated using multiple conventional assays (TPC, ABTS, FRAP, ORAC) and the oil-in-water (O/W) emulsion-based high throughput assays (the CAT and ApoCAT) in a comparison with an autoxidative O/W emulsion model monitored by formation of lipid hydroperoxide and TBARS values. Results suggest that only the ApoCAT assay is able to forecast the antioxidative performances of the extracts in O/W emulsions, regardless of the differences in extraction solvents, while the CAT assay can explain only the performance of ethanolic plant extracts in O/W emulsions. According to untargeted metabolite analysis, the antioxidants activities of plant extracts might be strongly influenced by extraction solvents. As a result, not only is the quantity of particular metabolites impacted, but also the whole metabolite (antioxidant) profiles of the extracts are modified. In addition, this study demonstrates that both the chemical reaction scheme and physical-state of a model are important parameters for designing a better antioxidant assay in the future.