To re-evaluate the various hypotheses on the systematic position of Parergodrilus heideri Reisinger, 1925 and Hrabeiella periglandulata Pizl & Chalupský, 1984, the sole truly terrestrial non-clitellate annelids known to date, their phylogenetic relationships were investigated using a data set of new 18S rDNA sequences of these and other five relevant annelid taxa, including an unknown species of Ctenodrilidae, as well as homologous sequences already available for 18 polychaetes, one aphanoneuran, 11 clitellates, two pogonophorans, one echiuran, one sipunculan, three molluscs and two arthropods. Two different alignments were constructed, according to an algorithmic method (Clustal W) and on the basis of a secondary structure model (DCSE). A maximum parsimony analysis was performed with arthropods as an unambiguous outgroup. With both alignments, the resulting topology confirms the validity of grouping P. heideri and Stygocapitella subterranea Knöllner, 1934 into the family Parergodrilidae. Hrabeiella periglandulata never clusters with them and its position relative to this and other polychaete families is still obscure, but a close relationship with aphanoneurans is suggested by the most parsimonious trees. All these taxa appear to be far from the Clitellata. Most relationships among polychaetes are not supported by significant bootstrap and Bremer values. These polytomies are corroborated by independent evidence and are interpreted as resulting from an ancient emergence and a rapid radiation of Polychaeta.