Determining which biological traits affect taxonomic durations is critical for explaining macroevolutionary patterns. Two approaches are commonly used to investigate the associations between traits and durations and/or extinction and origination rates: analyses of taxonomic occurrence patterns in the fossil record and comparative phylogenetic analyses, predominantly of extant taxa. By capitalizing upon the empirical record of past extinctions, paleontological data avoid some of the limitations of existing methods for inferring extinction and origination rates from molecular phylogenies. However, most paleontological studies of extinction selectivity have ignored phylogenetic relationships because there is a dearth of phylogenetic hypotheses for diverse non-vertebrate higher taxa in the fossil record. This omission inflates the degrees of freedom in statistical analyses and leaves open the possibility that observed associations are indirect, reflecting shared evolutionary history rather than the direct influence of particular traits on durations. Here we investigate global patterns of extinction selectivity in Devonian terebratulide brachiopods and compare the results of taxonomic vs. phylogenetic approaches. Regression models that assume independence among taxa provide support for a positive association between geographic range size and genus duration but do not indicate an association between body size and genus duration. Brownian motion models of trait evolution identify significant similarities in body size, range size, and duration among closely related terebratulide genera. We use phylogenetic regression to account for shared evolutionary history and find support for a significant positive association between range size and duration among terebratulides that is also phylogenetically structured. The estimated range size–duration relationship is moderately weaker in the phylogenetic analysis due to the down-weighting of closely related genera that were both broadly distributed and long lived; however, this change in slope is not statistically significant. These results provide evidence for the phylogenetic conservatism of organismal and emergent traits, yet also the general phylogenetic independence of the relationship between range size and duration.