Portunoidea is a diverse lineage of ecologically and economically important marine crabs comprising 8 families and 14 subfamilies. Closely related portunid subfamilies Caphyrinae and Thalamitinae constitute some of this group’s greatest morphological and taxonomic diversity, and are the only known lineages to include symbiotic taxa. Emergence of symbiosis in decapods remains poorly studied and portunoid crabs provide an interesting, but often overlooked example. Yet the paucity of molecular phylogenetic data available for Portunoidea makes it challenging to investigate the evolution and systematics of the group. Phylogenetic analyses, though limited, suggest that many putative portunoid taxa are para- or polyphyletic. Here I augment existing molecular data—significantly increasing taxon sampling of Caphyrinae, Thalamitinae, and several disparate portunoid lineages—to investigate the phylogenetic origin of symbiosis within Portunoidea and reevaluate higher- and lower-level portunoid classifications. Phylogenetic analyses were carried out on sequences of H3, 28S rRNA, 16S rRNA, and CO1 for up to 168 portunoid taxa; this included, for the first time, molecular data from the genera Atoportunus, Brusinia, Caphyra, Coelocarcinus, Gonioinfradens, Raymanninus, and Thalamonyx. Results support the placement of all symbiotic taxa (Caphyra, Lissocarcinus, and two Thalamita) in a single clade derived within the thalamitine genus Thalamita. Caphyrina Paulson, 1875, nom. trans. is recognized here as a subtribe within the subfamily Thalamitinae. Results also support the following taxonomic actions: Cronius is reclassified as a thalamitine genus; Thalamonyx is reestablished as a valid genus; Goniosupradens is raised to the generic rank; and three new genera (Zygita gen. nov., Thranita gen. nov., and Trierarchus gen. nov.) are described to accommodate some Thalamita s.l. taxa rendered paraphyletic by Caphyrina. A new diagnosis of Thalamitinae is provided. Results also support a more conservative classification of Portunoidea comprising three instead of eight extant families: Geryonidae (Geryonidae + Ovalipidae; new diagnosis provided), Carcinidae (Carcinidae + Pirimelidae + Polybiidae + Thiidae + Coelocarcinus; new diagnosis provided) and Portunidae. Finally, 16s rRNA data suggests family Brusiniidae might not be a portunoid lineage.