Increasing application of acoustic telemetry for determining survival, migration and habitat use of fishes highlights the need to improve interpretation of tracks that end abruptly: when is fishing mortality, predation, or some other cause to be inferred? Significant technological advances have led to the development of tags that “sense” predation and can be used to infer information about the type of predator that consumed the tagged fish. However, growing evidence suggests that bird predation is not effectively quantified by the technology. We hypothesized that reduction in sound transmission from acoustic tags in the gut of a bird combined with short bird diving intervals would eliminate detections of acoustic telemetry tags from the surface and severely reduce detection efficiency at depth. We test this hypothesis indirectly with two experiments using cormorant carcasses containing tagged fish in which carcasses were either tethered to a mooring for several hours or lowered through the water to simulate diving behavior. Detection of tagged prey fish in the gut of bird carcasses was severely reduced or negated completely, supporting our hypothesis. By comparison, as expected, tagged fish that were not in the gut of bird carcasses were detected at a higher frequency. Depth and distance to passive moored receivers also affected detection probability of tagged fish with more detections at depth and when closer to the receiver. Our results emphasized the importance of accounting for avian predation of tagged fish in studies of prey species in surface waters. Further, while recent development of predation sensing tags has illustrated a few examples of bird predation, our results demonstrate that determining that a tagged fish has been consumed by a diving bird will be difficult and will likely require alternative methods or technologies.