The importance of taxon sampling in phylogenetic accuracy is a topic of active debate. We investigated the role of taxon sampling in causing incongruent results between two recent phylogenomic studies of stinging wasps (Hymenoptera: Aculeata), a diverse lineage that includes ants, bees and the majority of eusocial insects. Using target enrichment of ultraconserved element (UCE) loci, we assembled the largest aculeate phylogenomic data set to date, sampling 854 loci from 187 taxa, including 30 out of 31 aculeate families, and a diversity of parasitoid outgroups. We analyzed the complete matrix using multiple analytical approaches, and also performed a series of taxon inclusion/exclusion experiments, in which we analyzed taxon sets identical to and slightly modified from the previous phylogenomic studies. Our results provide a highly supported phylogeny for virtually all aculeate lineages sampled, supporting ants as sister to Apoidea (bees+apoid wasps), bees as sister to Philanthinae+Pemphredoninae (lineages within a paraphyletic Crabronidae), Melittidae as sister to remaining bees, and paraphyly of cuckoo wasps (Chrysidoidea). Our divergence dating analyses estimate ages for aculeate lineages in close concordance with the fossil record. Our analyses also demonstrate that outgroup choice and taxon evenness can fundamentally impact topology and clade support in phylogenomic inference.