The effect of the composition of twelve varieties of extra virgin olive oils (EVOOs) on their differentiation based in agronomic criteria and on the antioxidant capacity was studied. Principal component analysis permitted an overview of the samples and their compositions, showing evidence of grouping and correlation between antioxidant capacity, oleuropein and ligstroside derivatives (OLD) and specific extinction at 270. Oleic and linoleic acids, 3,4-DHPEA-EA and p-HPEA-EDA (OLD), unsaturated/saturated ratio and induction time (IT) allowed the correct classification of samples according to year of harvest, ripening stage and variety. The antioxidant capacity of EVOOs was satisfactory predicted through a partial least square model based on ΔK, hydroxytyrosol, pinoresinol, oleuropein derivate and IT. Validation of the model gave a correlation R>0.83 and an error of 7% for independent samples. This model could be a useful tool for the olive industry to highlight the nutritional quality of EVOOs and improve their marketing.