Many fossil assemblages are time-averaged, with multiple generations of organisms mixed into a single stratigraphic horizon. A time-averaged sample of a taxon should be more variable than a single-generation sample if enough morphologic change occurred during the interval of time-averaging. Time-averaging may also alter correlations between morphologic variables and obscure allometric relationships in an evolving population. To investigate these issues, we estimated the variability of six modern, single-generation samples of the bivalve Mercenaria campechiensis using Procrustes analysis and compared them with several time-averaged Pleistocene samples of M. campechiensis and M. permagna. Both the modern and the fossil samples ranged in variability, but these ranges were virtually identical. Morphology was quite stable over the hundreds to many thousands of years that passed as the assemblages accumulated, and the variabilities of the fossil samples could be used to estimate single-generation variability. At one fossil locality, the environment and paleocommunity changed partway through the collection interval; the morphology of Mercenaria appears stable above and below the transition but changes across it. This change is similar in magnitude to the differences between geographically separated modern populations, whereas temporal variation within single environmental settings is distinctly less than geographic variation. Analytical time-averaging (the mixing of fossils from different horizons) between paleocommunities increased variability slightly (but not significantly) above that found in living populations. While its constituent populations appear stable on millennial timescales, M. campechiensis has been evolutionarily static since at least the early to middle Pleistocene.