Temporal frequency distributions of radiocarbon ages from archaeological sites can be used as a proxy record for human paleodemography after correction for taphonomic bias, or the time dependent loss of sediments due to erosion. Surovell et al. (2009) presented a global taphonomic correction model based on radiocarbon ages from volcanic deposits that has since been used by several researchers for paleodemographic reconstructions. This method is based on the assumption that the best indicator of relative human population density over time is not the absolute abundance of archaeological materials over time but, instead, the abundance of cultural material relative to geologic contexts in which those materials can occur. To verify the Surovell et al. model, in this paper we take 2457 radiocarbon ages from geologic contexts collected from published literature to create an independent model of taphonomic bias. We find that between 1 and 39 ka, the two curves are largely indistinguishable, but that they diverge in recent times. This suggests that current global models of taphonomic correction can be used to reconstruct human populations for the late Quaternary, but that demographic reconstructions remain challenging for the most recent two millennia.