Bulk samples, the key source of quantitative data in paleontology, are not sieved in a uniform way even for the same types of fossil materials, and there is no clear consensus among researchers regarding mesh size. Even when standard processing procedures have been published (e.g., for palynomorph recovery) there is variation in sieves used, depending on project goals. To test the effect of sieve size on quantitative data, we used a large dataset of Miocene mollusks from two adjacent bioprovinces (3870 specimens, 297 species), compiled from bulk samples and processed with fine sieves (<1mm mesh size). Using computer simulation the data were "sieved" at coarser levels, from 2mm to 10mm, at 1mm increments. Results show that taphonomic, paleoecological, and diversity patterns can all change dramatically, but not uniformly, as a function of sieve size: (1) drilling intensity increases with increasing sieve size, likely reflecting a concurrent decrease in the bivalve/gastropod ratio; (2) the proportion of right to left valves of bivalves remains relatively stable (~50:50 ratio) as mesh size increases; (3) the level of encrustation increases with coarser mesh sizes; and (4) the total taphonomic grade (a combination of many taphonomic characteristics) is highest at 5mm mesh size. These patterns do not necessarily remain consistent when data are analyzed separately for the two bioprovinces. Drilling intensity increases continuously, from 15% at 2mm mesh to nearly 27% at 10mm mesh, in one bioprovince (Boreal), but in the adjacent bioprovince (Paratethys) there is a decrease in drilling intensity from 12% at 2mm mesh to 7% at 6mm, followed by a slight increase to 10% at 10mm. Moreover, analyses of diversity and evenness patterns show that three dramatically different conclusions are suggested by data for sieve size of 2mm, 5mm, and 10mm, respectively. Thus even if the same mesh size is used consistently to create different datasets, outcomes of comparative analyses may change notably depending on what that mesh size is. A literature compilation indicates that mesh size varies greatly (from <1mm to 30mm) across quantitative studies. Achieving a methodological consensus is unlikely, but corrective measures for making samples processed with different sieves more comparable are feasible in some cases.