Genetic resiliency is the likelihood that populations retain sufficient genetic diversity to respond to environmental change. It is rarely examined through time in conservation genetic studies due to challenges of acquiring and sequencing historical specimens. Focusing on populations of two sibling bumble bee species of conservation concern with different recent patterns of decline, we used museum specimens collected between 1960 and 2020 and 15 microsatellite markers to assess genetic resiliency (allelic richness, expected heterozygosity, and inbreeding) through time and across geographic space. We find evidence of decreasing allelic richness through time, starting at least 30 years before observed abundance declines in one species and at least 20 years before present in a species with apparently stable abundance. We also found increasing expected heterozygosity through time, indicating increased inbreeding, in the putatively stable species. We demonstrate that genetic measurements taken from specimens collected through time can be used to detect population decline in imperiled species before decreases in abundance are detected. We also demonstrate the importance of interpreting population genetic metrics within the context of historical patterns to assess species' conservation statuses. Finally, we discuss the limitations of currently available population genetic methods, including the influence of isolation by distance and sampling density on measurements of genetic structure, and the influence of demographic characteristics and choice of genetic markers on estimates of genetic diversity and structure. We call for further development of individual-based modeling methods to measure genetic structure, as opposed to commonly applied population-based metrics, to overcome these limitations.