A new approach for evaluating the potential biases of walrus (Odobenus rosmarus) diet data derived from stomach contents was examined based on the rates at which different prey types were digested. In this study controlled digestion experiments on polychaetes, echiurid and sipunculid worms, clams, snails, and crabs demonstrated that these prey items did not remain equally identifiable during digestion. Polychaetes, echiurids, and sipunculids were the least persistent prey. All worms became unidentifiable during the six-hour digestion trials. Over 50% of the clams maintained their diagnostic tissues (i. e., foot and/or siphon). Clam viscera did not survive hour 2. Snails and crustaceans were the most persistent prey. Without a consideration of the state of digestion, a stomach sample may not accurately reflect the species composition and size of prey consumed. Volume is not a reliable measure of a prey's relative importance, as the diagnostic fragments of invertebrate prey in a stomach vary greatly in physical composition. Walruses probably consume all of the soft tissues of clams, not solely the foot and siphon.