Forecasting long-term biotic responses to anthropogenically-driven environmental change necessitates unlocking the wealth of baseline information contained in the recent fossil record. Homestead Cave and Two-ledges Chamber, long-term owl roosts in the Great Basin, contain stratified deposits of regurgitated pellets rich in small-mammal skeletal remains that record shifts in local faunal composition through the Holocene (past 10,000 years). In contrast to Two-ledges Chamber, the Homestead Cave record reveals a striking richness decline over the past ~60 years. To successfully use ecological information from such death-assemblage archives, however, the accuracy with which they record community composition must be quantified, thus testing the strength of the biological signals they capture. Here I compare modern trapping surveys, modern “pellet rain”, and Holocene death-assemblages from both caves to assess fidelity and recent community shifts. Preliminary results for Homestead Cave indicate high rank-order correspondence among modern and fossil death-assemblages and the living community, which weakens as the amount of summed time and elapsed time between samples increases (Spearman Rho ~0.9-0.3). Additionally, modern and fossil death-assemblages are richer than snapshot trapping surveys, with 60-100% more taxa. Decreasing live-dead correspondence, coupled with declining richness, suggests a community in transition, thus analyses are underway to differentiate the effects of trap bias and predator selectivity as potential confounding factors. In summary, death-assemblages represent valuable archives of biodiversity information. In addition, live-dead analyses can highlight taxa with unexpected offsets in the living and the dead. Thus combining “dead data” with modern surveys can reveal ecosystem transitions and highlight particularly sensitive taxa.