Evolutionary change during the interval in which a fossil sample accumulates will inflate the variance of that sample relative to the population-level standing variation. If this effect is widespread and severe, paleontological samples will not provide reliable estimates of population variation. Although the few published studies conducted to test this possibility have found similar levels of variation in samples differing greatly in temporal acuity, the paucity of case studies prevents assessing the generality of this pattern. In this paper, two independent, literature-based approaches are used to greatly expand the data available to address this issue. The first approach compares morphometric variability in Quaternary mammal samples with samples from related modern populations. The second approach artificially lumps separate samples from evolving lineages and calculates the variance effects of this analytical time-averaging. Both approaches yield consistent results indicating that variance observed in time-averaged samples is typically only slightly inflated (approximately 5%) relative to population-level values. This finding suggests that rates of evolution are typically slow when scaled to within-population variation, providing support for relative stasis as the dominant mode of within-lineage evolution. An important practical consequence of these findings is that time-averaged fossil samples generally show trait variances and covariances that are similar to population-level parameters, which has been an important but implicit assumption in many paleontological studies of phenotypic variation.