Predatory drill holes in marine invertebrates are an important source of information on the nature of biotic interactions and are often used to explore the ecological and evolutionary roles of such interactions in deep time. Measures of drilling frequencies and drill-hole site stereotypy represent the raw data for inferring the intensity and selectivity of drilling predation. One potential source of bias explored in this study relates to the hydrodynamic properties of shells: presence of drill holes and/or drill-hole position may influence how shells behave when subjected to moving fluids. In a unidirectional flow-tank study with the bivalve Donax scortum Linnaeus, 1758, we found that the threshold current velocity for the entrainment of undrilled convex-up shells is significantly lower than for centrally drilled shells, which could be explained by Bernoulli's principle. The position of the drill hole on a shell also affects its hydrodynamic properties because umbonally drilled shells require a lower entrainment velocity than centrally drilled shells. This difference could potentially result in assemblages of different stereotypic patterns. We also demonstrate the extent of alteration of an assemblage by this process using a simulation parameterized with experimental results. The latter show that a dramatic change in inferred drilling intensity, size selectivity, and stereotypic patterns from the original population can be produced by hydrodynamic sorting. Our study indicates that such sorting can yield a sample significantly different from the original one in terms of drill-hole characteristics. Hence, the effect of such bias should be assessed before inferring the nature of biotic interaction of fossil assemblages.