Nomenclatural stability is a widely stated objective of taxonomy, but the Linnaean binomial system is not well suited to promote stability of species names. Nor does the binomial system provide adequate flexibility to cope with unclear generic boundaries. Caryopteris (Lamiaceae) is used to illustrate these problems. Cladistic analyses of morphological and molecular data indicate that Caryopteris is either paraphyletic or polyphyletic, but resolution is too poor to assign all 16 species to well supported clades. Ideally, one would like to name the well supported clades as genera and designate the remaining species that are currently included in Caryopteris as "incertae sedis" within subfamily Ajugoideae. However, the binomial system cannot accommodate uncertainty about genus-level relationships. This forces one to choose between unsatisfactory alternative classifications that either conflict with the best estimate of phylogeny or require naming several new, marginally distinct monotypic genera. Whichever approach is chosen, correcting the current, patently unacceptable circumscription of Caryopteris will require new combinations for at least nine species. An alternative form of species name, a hyphenated uninomial in which the word before the hyphen is not necessarily the current genus assignment, would be much more stable and would facilitate changes in genus-level classification. In the Caryopteris example, the new understanding of relationships could be translated into classification without any name changes at all. Transition to a system of hyphenated uninomials would not be traumatic if the uninomials were based on currently accepted binomials and utilized the same types. Hyphenated uninomials would be compatible with both the rank-based system of nomenclature represented by the ICBN and the rankless phylogenetic nomenclature proposed by de Queiroz and Gauthier.

Key words: binomial nomenclature, Caryopteris, Labiatae, Lamiaceae, nomenclatural stability, phylogeny