Phylogenetic relationships within the Sphacelariales sensu stricto were investigated using both a molecular and a morphological approach. Twenty species were included, representing all three families (i.e. Cladostephaceae, Sphacelariaceae and Stypocaulaceae) and six of the eight genera. The outgroup consisted of six species representing the Syringodermatales, Dictyotales, Choristocarpaceae and Onslowiaceae. DNA sequences of partial rbcL [1255–1375 nucleotides (nt)] and the adjacent RUBISCO spacer (40–842 nt) were determined in order to assess the molecular phylogeny. Only the 3′-end of the spacer (112 nt positions) was alignable for the ingroup taxa. Partition homogeneity testing showed that rbcL and RUBISCO spacer sequences could be combined. However, analysis of rbcL alone or in combination with the RUBISCO spacer gave the same results with only slight differences in support (bootstrap, jackknife, decay). Support was low at the base of the ingroup. Four basal clades could be discerned: (1) Stypocaulaceae, (2) Sphacelaria radicans, (3) Sphacelaria caespitula and (4) all other included taxa: (Sphacella subtilissima+the Sphacelaria subgenus Propagulifera) and (Cladostephus spongiosus+Sphacelaria nana+the Sphacelaria subgenera Battersia+Pseudochaetopteris). The independent morphological analysis (using 23 unordered morphological characters) revealed high homoplasy and an almost completely unresolved tree in which only the subgenus Propagulifera was supported. The morphological characters were subsequently mapped onto the rbcL tree in order to identify diagnostic or phylogenetically informative characters. Blackening in response to bleach and the presence of secondary segments were found to be basal synapomorphies for the Sphacelariales; presence of propagules with a lenticular central apical cell defines the Propagulifera; and strict acroblastic branching mode and axillary zoidangia define the Stypocaulaceae. The remaining characters have been gained or lost multiple times. This study highlights the problem of extreme morphological convergence and/or plasticity. Four options for a new circumscription including nomenclatural changes are discussed. It is concluded that none of the options will lead to greater clarity with respect to either identification or classification.