Abstract-The delayed mortality rate of crab discarded during fishing operations can be under-or overestimated in laboratory holding experiments, given the unnatural conditions and the short-term duration of these experiments. To evaluate the extent to which a method affects accuracy in these estimations, we compared mortality rates established through laboratory holding with mortality inferred from a yearand-a-half long tag-return study of Dungeness crab (Cancer magister) discarded in Oregon crab fisheries. The reflex action mortality predictor (RAMP) approach, which relates reflex impairment to probability of mortality, was applied in both studies. Similar patterns in mortalityand tag return-rates with respect to fishery, sex, reflex impairment, shell hardness, and injury from the 2 studies lends support to the reliability of the laboratory-generated mortality rates. However, results suggest that mortality rates determined in captivity are likely underestimated when crab are dropped a distance of greater than 6 m (and potentially less) back to water. This underscores the importance of determining the contribution to mortality of variables in the capture, handling, and discard process that are not incorporated in a study to estimate mortality of discarded animals. Both studies also highlighted the significance of sample size when applying the RAMP approach to a fishery with low rates of discard mortality.