Using measurements of naturally occurring stable isotopes to reconstruct diets or source of feeding requires quantifying isotopic discrimination factors or the relationships between isotope ratios in food and in consumer tissues. Diet-tissue discrimination factors of carbon ((13)C/(12)C, or delta (13)C) and nitrogen ((15)N/(14)N, or delta (15)N) isotopes in whole blood and feathers, representing noninvasive sampling techniques, were examined using three species of captive penguins (king Aptenodytes patagonicus, gentoo Pygoscelis papua, and rockhopper Eudyptes chrysocome penguins) fed known diets. King and rockhopper penguins raised on a constant diet of herring and capelin, respectively, had tissues enriched in (15)N compared to fish, with discrimination factors being higher in feathers than in blood. These data, together with previous works, allowed us to calculate average discrimination factors for (15)N between whole lipid-free prey and blood and feathers of piscivorous birds; they amount to +2.7 per thousand and +4.2 per thousand, respectively. Both fish species were segregated by their delta (13)C and delta (15)N values, and importantly, lipid-free fish muscle tissue was consistently depleted in (13)C and enriched in (15)N compared to whole lipid-free fish. This finding has important implications because previous studies usually base dietary reconstructions on muscle of prey rather than on whole prey items consumed by the predator. We tested the effect of these differences using mass balance calculations to the quantification of food sources of gentoo penguins that had a mixed diet. Modeling indicated correct estimates when using the isotopic signature of whole fish (muscle) and the discrimination factors between whole fish (muscle) and penguin blood. Conversely, the use of isotopic signatures of muscle together with discrimination factors between whole fish and blood (or the reverse) leads to spurious estimates in food proportions. Consequently, great care must be taken in the choice of isotopic discrimination factors to apply to wild species for which no controlled experiments on captive individuals have been done. Finally, our results also indicate that there is no need to remove lipids before isotopic analysis of avian blood.