A phylogenetic analysis of species potentially belonging to the Tanytarsus eminulus, gregarius, mendax and lugens species groups is performed using morphological characters from the adult male, pupa and larva. The results show that morphological characters do not support the postulated monophyly of the eminulus, gregarius, lugens and mendax group combined in unweighted parsimony analyses, and that a constraint based on unique synapomorphies and evidence from molecular data have to be used in order to produce cladograms with reasonable topologies. Four reasons for this are discussed: Few taxa, few characters, choice of secondary outgroup taxa and a high amount of homoplasy in the data set. A hierarchial analysis procedure is used to avoid the numerous question marks in the complete data matrix. In the preferred tree, the traditional species groups within Tanytarsus are kept, and one new species group, the mcmillani group, comprising only old Gondwanan species is erected. The results are compared to other recent studies on chironomid species relationships, and comments are given to the zoogeographical patterns of the species in the eminulus, gregarius, lugens, mcmillani and mendax species groups.