From a traditional ‘Volterra-Gause’ point of view, populations and species compete for resources on local scales and such interactions influence or determine the makeup of local assemblages. Whether or not this simple version of competition can be scaled up to evolutionary time in order to explain taxonomic patterns in the history of life continues to be debated. Here I examine the empirical foundation for competition-driven diversity trends. Results are based on comparative analyses of 30 large-scale 'competitive displacements' among species-rich fossil groups, culled from the paleontological literature of the last 50 years. I compared authors’ treatments of defining competition (resource- or outcome-based), ecological equivalence, reliance on uniformitarianism, logical domains, units to be counted (taxon rank, sufficiency as species proxies), groups compared (clades, grades, guilds), methods of analysis (by interval, by assemblage), and whether or not validation was supported by congruent patterns of abundance, biogeography, or morphology. Most empirical studies to date: 1) rely principally on inductive and uniformitarianist reasoning, 2) fail to justify ecological equivalence of putative competitors or presumed mechanisms of competition, 3) are based on family ranks and do not consider the sufficiency of families as species proxies, 4) are summed by geologic interval and thus do not account for co-occurrence patterns within assemblages, and 5) infer taxonomic turnover principally in terms of increased extinction of the displaced group without considering origination dynamics. Geographic displacement or abundance shifts are seldom documented, but when present can provide strong independent validation of a competition hypothesis or evidence for alternative causal explanations. Predictive models (particularly those based on coupled-logistic equations) hold promise but remain at an early developmental stage, limited by difficulties in estimating rate parameters and by problematic assumptions of equilibria. Ecological studies over the past decade have shown that many if not most local communities are unsaturated. Coupled with molecular data showing sparse populations of species over broad geographic areas, these studies may imply that simple equilibrial assumptions are not justified.