We assessed phylogenetic relationships within Serpulidae (including Spirorbinae) using parsimony and Bayesian analyses of 18S rDNA, the D1 and D9-D10 regions of 28S rDNA, and 38 morphological characters. In total, 857 parsimony informative characters were used for 31 terminals, 29 serpulids and sabellid and sabellariid outgroups. Following ILD assessment the two sequence partitions and morphology were analysed separately and in combination. The morphological parsimony analysis was congruent with the results of the 2003 preliminary analysis by Kupriyanova in suggesting that a monophyletic Serpulinae and Spirorbinae form a clade, while the remaining serpulids form a basal grade comprising what are normally regarded as Filograninae. Bremer support values were, however, quite low throughout. In contrast, the combined analyses of molecular and morphological data sets provided highly resolved and well-supported trees, though with some conflict when compared to the morphology-only analysis. Spirorbinae was recovered as a sister group to a monophyletic group comprising both 'filogranin' taxa (Salmacina, Filograna, Protis, and Protula) and 'serpulin' taxa such as Chitinopoma, Metavermilia, and Vermiliopsis. Thus the traditionally formulated subfamilies Serpulinae and Filograninae are not monophyletic. This indicates that a major revision of serpulid taxonomy is needed at the more inclusive taxonomic levels. We refrain from doing so based on the present analyses since we feel that further taxon sampling and molecular sequencing are required. The evolution of features such as the operculum and larval development are discussed.