Groups of rainbow trout (Salmo gairdneri) of approximate mean initial weight 8 g were grown in outdoor tanks over a 14-week period at water temperatures between 12 degrees (start) and 6 degrees (end). Four diets were used. Two contained non-oxidized fish oil (120 g/kg) with or without supplementary DL-alpha tocopheryl acetate and two contained moderately oxidized fish oil again with or without DL-alpha-tocopheryl acetate. The measured selenium content of the diets was 0.10 mg/kg. No significant differences occurred as a consequence of the use of moderately oxidized oil compared with the corresponding treatments using non-oxidized oil. Significant differences did occur between dietary treatments that contained supplementary DL-alpha-tocopheryl acetate and those that did not. These differences applied to weight gain, haematocrit, erythrocyte fragility, mortalities, liver and muscle tocopherol concentrations and lipid peroxidation of liver mitochondria in vitro. Liver glutathione peroxidase (EC 1.11.1.9) activity was unaffected by the dietary treatments used and the proportions of fatty acids in polar lipids of liver and muscle were little changed by the diets used. Severe muscle damage occurred in trout given diets lacking supplementary DL-alpha-tocopheryl acetate. Previous experiments carried out on rainbow trout at a constant water temperature of 15 degrees ( Hung et al. 1981; Cowey et al. 1981, 1983), using diets lacking supplementary vitamin E, did not lead to differences in weight gain, pathological changes or mortalities. Vitamin E requirement may increase as water temperature decreases; minimum dietary requirements for vitamin E measured at a constant water temperature of 15 degrees may not be valid under practical conditions where water temperatures vary over the year.