The phylogenetic relationships of the Clitellata were investigated with a data set of published and new complete 18S rRNA gene sequences of 51 species representing 41 families. Sequences were aligned on the basis of a secondary structure model and analysed with maximum parsimony and maximum likelihood. In contrast to the latter method, parsimony did not recover the monophyly of Clitellata. However, a close scrutiny of the data suggested a spurious attraction between some polychaetes and clitellates. As a rule, molecular trees are closely aligned with morphology-based phylogenies. Acanthobdellida and Euhirudinea were reconciled in their traditional Hirudinea clade and were included in the Oligochaeta with the Branchiobdellida via the Lumbriculidae as a possible link between the two assemblages. While the 18S gene yielded a meaningful historical signal for determining relationships within clitellates, the exact position of Hirudinea and Branchiobdellida within oligochaetes remained unresolved. The lack of phylogenetic signal is interpreted as evidence for a rapid radiation of these taxa. The placement of Clitellata within the Polychaeta remained unresolved. The biological reality of polytomies within annelids is suggested and supports the hypothesis of an extremely ancient radiation of polychaetes and emergence of clitellates.