Paleoecological studies requiring taxonomic information beyond presence/absence data (i.e. relative abundance) must demonstrate that the sampling regime employed to collect such data adequately fits the faunas under investigation. A study of dense shell beds of the Middle Eocene Gosport Sand of Alabama indicates that, for localities with high abundance and diversity and little internal structure, diversity-abundance data derived from standard-sized samples vary significantly and cannot be used to develop consistent species-level distributional patterns. Based on this result and the additional observation that different species show different patterns of occurrence, we propose an operational approach for determining what sample size will yield reliable distributional data for different proportions of the fauna. In the Gospon, samples of 3–5***1 yield reliable data on the approximately 25% of the species that are most common and abundant; samples of 35–40 I would be necessary to yield similarly reliable data on the next 15-10% the remainder of species are too rare ever to be realistically examined quantitatively in bulk samples. □Sampling, paleoecology, taphonomy, Mollusks, Gosport Sand, Eocene.