Lipid oxidation is a major factor in meat quality. In order to relate human perceptions of lipid oxidation, as determined by a trained taste panel, to a chemical measurement of oxidation, we studied meat from animals with a wide range of potential oxidation through differences in their PUFA composition and by displaying the meat in high oxygen modified atmosphere packs for varying lengths of time. Meat was obtained from 73 Angus- and Charolais-cross steers from different trials that had been raised on 10 different diets: grass silage (high in C18:3, n-3), cereal concentrate (high in C18:2, n-6), three diets with 3% added fat consisting of three levels of protected lipid supplement (high in C18:2, n-6 and C18:3, n-3, ratio 1:1), a control with Megalac(®) (relatively saturated), three diets with three levels of inclusion of protected fish oil (high in C20:5 n-3 and C22:6 n-3) plus a constant amount of unprotected fish oil and a final diet with an unprotected fish oil control. The longissimus dorsi muscle was excised from the left carcass side, aged vacuum packaged for 10-13 days depending on the projects and frozen for less than eight months. TBARS and sensory analyses were performed on steaks displayed for 0, 4 or 9 days under simulated retail conditions, exposed to light in modified atmosphere packaging (CO(2):O(2); 25:75). Meat oxidation increased throughout display for each of the diets, as shown by a rise in TBARS values. This increase was not linear, differences between 0 and 4 days of display were smaller than between 4 and 9 days of display. The lowest TBARS and lowest increment occurred in the two control diets and the grass-fed animals, probably due to the more saturated fat of meat from animals fed the control diets and the higher content of vitamin E. Sensory attributes were also influenced by time of display. Positive attributes, such as beef flavour or overall liking, decreased throughout display, whereas negative attributes, such as abnormal and rancid flavours, increased. The correlations between sensory and analytical attributes were high. TBARS were a good predictor of the perception of rancidity (Spearman's rho=0.84). Panellist preferences were related to the presence of beef flavour (rho=0.93) and to the absence of abnormal (rho=-0.88) and rancid flavours (rho=-0.83). Under the experimental conditions used, a TBARS value of around 2 could be considered the limiting threshold for the acceptability of oxidised beef.