Natural antioxidants such as tocopherols, ascorbic acid, carotenoids, flavonoids can delay lipid oxidation and are widely used for that purpose in foods. Their effect can be highlighted thanks to different in vitro methods useful to clarify the mechanism of action of a molecule or a family of molecules. Unfortunately, most of these methods are not predictive for food systems. In this study, we wanted to better understand mechanisms occurring in different food systems. For that, we decided to focus on onion extracts processed by different technologies. First, all nine extracts were analyzed to determine composition in total phenolic compounds by Folin Ciocalteu method, quercetins, glucoside quercetins and tocopherols. Then their chemical reactivity as antioxidants was evaluated with a pool of different methods. Ferric reducing antioxidant power (FRAP), DPPH° assay and the capacity to chelate ferrous iron were determined for each extract to select the most effective one to be tested in different food matrices such as bulk oil, emulsions and mayonnaise. In bulk oil, no effect on the delay of oxidation was observed whatever the tested onion extract. However, two extracts resulted in a lower peroxide value at end point (35 days) with peroxide value ranging from 52 to 58 MEq O2/ kg Oil compared to 77 MEq O2/ kg Oil for the oil without any antioxidant. In emulsions, a good effect of onion extracts standardized in quercetin aglycone was observed in comparison with quercetin alone. When oxidation was induced by Ferrous ion addition, 3 groups of antioxidants could be discriminated according to their efficiency at 1000 ppm. The most efficient extract in the non-iron induced emulsion becomes the less efficient in the induced emulsion. Water-extracted samples seem to be more effective in this induced model. Analysis of primary and secondary oxidation compounds enabled us to select three extracts that appeared to have interesting antioxidant behavior; they were then tested in a r