Soybean, cottonseed, corn and squash seed oils were treated for 3 h with activated carbon (l%, 2.5%, and 5%) in a batch adsorption system at 50 and 90°C. Peroxides (P,) and carotenoids (C,) were determined after adsorption. Equations that described the P, and C, variability were developed through discriminatory multiple regression analysis using as independent variables chemical characteristics of the oils (iodine and saponification index, peroxides, carotenoids and free fatty acids, saturated/unsaturated fatty acids ratio), process temperature and adsorbent concentration. Variables in the models explained 185% and =93% of the total variability observed in the P, and C, concentrations respectively, independent of oil source.