The graptolites, known only from fossils, have been convincingly allied to the pterobranch hemichordates, a group of tiny, mostly colonial marine invertebrates bearing feeding arms. The phylogenetic position of pterobranchs has been subject to debate and revision for over a century. Their colonial lifestyle and feeding arms were originally seen as evidence placing them among the bryozoans, until later and more careful anatomical studies revealed more characters in common with acorn worms. Pterobranchs and acorn worms are now grouped as the phylum Hemichordata. For many decades, it was thought that pterobranchs were closer to the ancestral form of hemichordates, particularly because ‘lophophorate’ invertebrates also possess feeding arms, notably the phoronids and bryozoans, as do crinoid echinoderms. This traditional view has been challenged by recent molecular evidence. First, there is strong molecular evidence to indicate that lophophorates are very distant from hemichordates and echinoderms, in a different major branch of the animal phylogenetic tree. Therefore, similarities between the feeding structures must be due to convergent evolution. Second, there is strong evidence that hemichordates and echinoderms form a clade (Ambulacraria) within the deuterostomes, rather than hemichordates being closer to chordates. Third, there is weaker evidence that pterobranchs may be derived from acorn worms, and hence that the vermiform body plan may be ancestral within hemichordates. This suggestion warrants further testing. Here we review the evidence for these conclusions, highlight strengths and weaknesses in the data and analyses, and consider the implications for the origins of pterobranchs and graptolites.