Natural accumulations of skeletal remains represent a valuable source of ecological data for paleontologists and neontologists alike. Use of these records requires a quantitative assessment of the degree to which potential biasing factors affect how accurately ecological information from the living community is recorded in the sedimentary record. This has been a major focus in recent years for taphonomists working with marine records, yet terrestrial systems have remained virtually unstudied—particularly communities of small-bodied taxa. Our ability to assess the potential origins and effects of postmortem bias in terrestrial skeletal assemblages (both modern and fossil) has therefore been limited. Predation is a common mechanism by which small-mammal skeletal remains are concentrated; raptors regurgitate the remains of their small-mammal prey in pellets rich in skeletal material, which accumulate below long-term roosting sites, especially in protected areas such as caves and rock shelters. Here I compare small-mammal death assemblages concentrated via owl predation at Two Ledges Chamber, a long-term owl cave roost in northwestern Nevada, with data from modern trapping surveys to evaluate (1) their ecological fidelity to the modern small-mammal community, (2) the effects of temporal variation and time-averaging (over months to centuries) on live-dead agreement, and (3) how spatial averaging affects the landscape-scale picture of the small-mammal community as reconstructed from dead remains. Despite potential obstacles to the recovery of ecological information from skeletal deposits generated via predation, I find high live-dead agreement across all ecological metrics and all temporal comparisons. I also find that the effects of time-averaging (specifically increased species richness of the death assemblage) become significant only at the century scale. Finally, I combine a mixing model approach with a principal coordinates analysis to show that the owls at Two Ledges Chamber sample from all habitats present in the immediate vicinity of the cave, producing a high-fidelity snapshot of the community that is spatially integrated at the local landscape scale.