The most common method for estimating adult survival in site specific demographic studies of Neotropical-Nearctic migratory bird populations is by measuring the return rate of marked individuals. Return rate historically has been defined as the ratio of resighted birds to the total number banded (i.e., with bands on) the prior year, and has been used as a 'minimum number known alive' estimate of survival. Return rates potentially underestimate true survival (the complement of mortality) for two reasons. First, not every bird that returns to the study plot is actually encountered by the field researcher, and second, not every bird that survives to the next breeding season returns to the study plot. We use a branching-tree diagram to illustrate that the essential problem with return rate methodology is that the fate of birds that are not resighted is unknown. It is widely recognized that Cormack-Jolly-Seber based analyses greatly improve 'survival' estimates by incorporating the probability of resighting a bird given that it is alive and present on the study plot. However, these estimates will underestimate true survival if birds disperse beyond the range of the resighting effort. Because long-distance dispersal events are an important component of migratory bird ecology, we cannot estimate true survival from return rate data until better information on dispersal distances and probabilities are collected. We discuss several conservation implications of underestimating survival,.and suggest terminology that is potentially less confusing.