Many species of large mammals were driven to extinction during the late Pleistocene and early Holocene (approx. 10,000 - 50,000 years ago), with cascading effects on the physical structure of ecosystems and the dispersal of seeds, nutrients, and microbes. However, it remains uncertain whether the parasites associated with these extinct hosts also disappeared or persisted in surviving (extant) mammals. We hypothesize that if some parasites endured, extant mammals sharing their ranges with phylogenetically similar extinct mammals would have a greater pathogen richness than expected based on current levels of host diversity. We find that the inclusion of variables related to these extinctions account for an additional 5% of deviance when modelling per-host viral and bacterial richness, compared to models run without these variables. Viral and bacterial richness for extant mammals was calculated from host-pathogen associations within the CLOVER repository. Partial dependence plots show a positive correlation between the number of extinct mammals lost and per-host viral and bacterial richness (p < 0.001 and p = 0.03, respectively). Additionally, decreasing phylogenetic distance between the extinct and extant species is associated with an increasing viral richness (p < 0.001). We discuss four mechanisms that may be driving these patterns and highlight future research to distinguish between them. Next, we use the models and IUCN range maps to identify geographic regions where viral and bacterial richness differs due to the inclusion of extinction variables. Notably, the richness of both pathogen types is increased in South America (viruses: +6.8%; bacteria: +3.1%) and decreased in Africa (viruses: -2.6%; bacteria: -13.6%), two continents known to have high and low levels of historical mammal extinctions, respectively. Viral richness is also elevated in North America (+8.6%), Europe (+5.1%), Oceania (+3.3%), and Asia (+2.3%). These results support the inclusion of extinction variables in future models of pathogen richness and may allow for improved targeting of future surveillance efforts.