We investigated the use of stable-carbon isotope analysis of serum and cellular fractions of blood to detect the extent of terrestrial feeding in polar bears on land during the ice-free period in western Hudson Bay. We compared blood in bears that were restricted entirely to coastal areas, who showed no evidence of terrestrial feeding, with blood in bears sampled at inland locations and who were known to have fed on berries of Vaccinium uliginosum and Empetrum nigrum. Despite a separation of approximately 9‰ between terrestrial and marine foods, we found no statistical difference in blood 613C values between these two groups of bears. This suggests that (1) carbon pathways associated with feeding on berries result in minor incorporation of terrestrial-based carbon into bulk plasma or cellular fractions of blood, (2) bears feed insignificantly on berries despite observational evidence to the contrary, or (3) carbon mobilized from endogenous lipid reserves overwhelmed the terrestrial signal or could not be segregated isotopically from carbon derived from berry carbohydrates. We discuss evidence for each of these scenarios and suggest that a more effective approach to using stable-carbon isotope analysis to delineate the importance or use of terrestrial foods to polar bears on land in Hudson Bay during the ice-free period might be through the isotopic analysis of exhaled carbon dioxide rather than blood components.