The study of parasite evolution relies on the identification of free-living sister taxa of parasitic lineages. Most lineages of parasitic helminths are characterized by an amazing diversity of species that complicates the resolution of phylogenetic relationships. Acanthocephalans offer a potential model system to test various long-standing hypotheses and generalizations regarding the evolution of parasitism in metazoans. The entirely parasitic Acanthocephala have a diversity of species that is manageable with regards to constructing global phylogenetic hypotheses, exhibit variation in hosts and habitats, and are hypothesized to have close phylogenetic affinities to the predominately free-living Rotifera. In this paper, I review and test previous hypotheses of acanthocephalan phylogenetic relationships with analyses of the available 18S rRNA sequence database. Maximum-parsimony and maximum-likelihood inferred trees differ significantly with regard to relationships among acanthocephalans and rotifers. Maximum-parsimony analysis results in a paraphyletic Rotifera, placing a long-branched bdelloid rotifer as the sister taxon of Acanthocephala. Maximum-likelihood analysis results in a monophyletic Rotifera. The difference between the two optimality criteria is attributed to long-branch attraction. The two analyses are congruent in terms of relationships within Acanthocephala. The three sampled classes are monophyletic, and the Archiacanthocephala is the sister taxon of a Palaeacanthocephala Eoacanthocephala clade. The phylogenetic hypothesis is used to assess the evolution of host and habitat preferences. Acanthocephalan lineages have exhibited multiple radiations into terrestrial habitats and bird and mammal definitive hosts from ancestral aquatic habitats and fish definitive hosts, while exhibiting phylogenetic conservatism in the type of arthropod intermediate host utilized.