During the last glacial period, several large carnivore species went extinct in Europe. Both abrupt and gradual climate changes have been blamed for these disappearances, but fossils must be precisely dated to establish whether their presence correlates with climatic events. Radiocarbon dating is a potentially powerful tool, as it directly dates faunal bone and can produce dates up to 50,000 BP. However, current radiocarbon datasets for carnivore species are small and of incomplete geographical coverage. This is partly due to the challenges of dating degraded bone collagen, which is often found in southern Europe. Furthermore, many Pleistocene radiocarbon dates likely underestimate the sample's true age, and dates often come with errors too large to match to high-resolution climatic events with confidence. In this paper, we will show that these weaknesses can be addressed by using robust pretreatment methods. In addition to ultrafiltration, routinely used in the Oxford Radiocarbon Accelerator Unit since 2000, novel compound-specific methods are being used to improve dating accuracy and expand the ability to date poorly-preserved samples. Bones of cave lion, panther, and hyena in southern Europe have been targeted for this study, as these carnivore species have few direct dates and collagen preservation in the region is problematic. When combined into a GIS model with climate records, these new results improve our understanding of how carnivores responded to climate change.