Mass extinctions can have dramatic effects on the trajectory of life, but in some cases the effects can be relatively small even when extinction rates are high. For example, the Late Ordovician mass extinction is the second most severe in terms of the proportion of genera eliminated, yet is noted for the lack of ecological consequences and shifts in clade dominance. By comparison, the end-Cretaceous mass extinction was less severe but eliminated several major clades while some rare surviving clades diversified in the Paleogene. This disconnect may be better understood by incorporating the phylogenetic relatedness of taxa into studies of mass extinctions, as the factors driving extinction and recovery are thought to be phylogenetically conserved and should therefore promote both origination and extinction of closely related taxa. Here, we test whether there was phylogenetic selectivity in extinction and origination using brachiopod genera from the Middle Ordovician through the Devonian. Using an index of taxonomic clustering (RCL) as a proxy for phylogenetic clustering, we find that A) both extinctions and originations shift from taxonomically random or weakly clustered within families in the Ordovician to strongly clustered in the Silurian and Devonian, beginning with the recovery following the Late Ordovician mass extinction, and B) the Late Ordovician mass extinction was itself only weakly clustered. Both results stand in stark contrast to Cretaceous-Cenozoic bivalves, which showed significant levels of taxonomic clustering of extinctions in the Cretaceous, including strong clustering in the mass extinction, but taxonomically random extinctions in the Cenozoic. The contrasting patterns between the Late Ordovician and end-Cretaceous events suggest a complex relationship between the phylogenetic selectivity of mass extinctions and the long-term phylogenetic signal in origination and extinction patterns.