During the last decade, much progress have been made on elucidating the phylogeny of Rosaceae through the use of molecular sequence data. David Morgan and coworkers (1994) found support for a re-circumscribed Rosoideae clade. We analysed Rosoideae in this strict sense, mainly using ITS and trnL/F sequence data, rooting trees on the Filipendula branch. Most newly discovered clades of the "backbone" of the Rosoideae tree are now well supported, and some of these require names. Other well supported clades include: Colurieae with Fallugia, Sieversia, Waldsteinia and Geum s. lat.; Rosa; Sanguisorbeae with the petaloid Agrimonia, Aremonia, Leucosidea, and Hagenia, as sister to a non-petaloid clade containing Sanguisorba, Acaena, Polylepis, among others; Potentilleae divided into Potentilla s. str. (including P. anserina and relatives) which is sister to Fragariinae, the latter containing Alchemilla, Fragaria, Chamaerhodos, Comarum, Sibbaldiopsis, some Sibbaldia as well as at least some Potentilla species. In our sample, Acaena, Geum (and most of its segregates), Sibbaldia, and Potentilla, as delimited traditionally, are polyphyletic. The position of Rubus is not securely resolved. If the remnants of Dryadeae is sister to Rosoideae, as some analyses indicate, the wind dispersed achenes with elongate plumose styles may be plesiomorphic within Rosoideae. The similarity of the fruits of some species in Colurieae and Dryadeae would then be explained, but it might make the interpretation of the evolution of the Filipendula fruit more complex.

Key words: Phylogeny Rosoideae Rosaceae