Resolving the global phylogeny of eukaryotes has proven to be challenging. Among the eukaryotic groups of uncertain phylogenetic position are jakobids, a group of bacterivorous flagellates that possess the most bacteria-like mitochondrial genomes known [1, 2]. Jakobids share several ultrastructural features with malawimonads and an assemblage of anaerobic protists (e.g., diplomonads and oxymonads) [3, 4]. These lineages together with Euglenozoa and Heterolobosea have collectively been designated “excavates” [5]. However, published molecular phylogenies based on the sequences of nuclear rRNAs [5–7] and up to six nucleus-encoded proteins [8–10] do not provide convincing support for the monophyly of excavates, nor do they uncover their relationship to other major eukaryotic groups [5–10]. Here, we report the first large-scale eukaryotic phylogeny, inferred from 143 nucleus-encoded proteins comprising 31,604 amino acid positions, that includes jakobids, malawimonads and cercozoans [7]. We obtain compelling support for the monophyly of jakobids, Euglenozoa plus Heterolobosea (JEH group), and for the association of cercozoans with stramenopiles plus alveolates. Furthermore, we observe a sister-group relationship between the JEH group and malawimonads after removing fast-evolving species from the dataset. We discuss the implications of these results for the concept of “excavates” and for the elucidation of eukaryotic phylogeny in general.