The primitive woody Angiosperm family Winteraceae is centred in the southwest Pacific, has an outpost in Madagascar (1 species), and a section of Drimys (4 species) in the New World. Pseudowintera is restricted to New Zealand, the Old World section Tasmannia of Drimys extends from the Philippines to Tasmania. Since A. C. Smith’s review of the family in 1943 about ten times as much material has become available and it appears that the taxonomy of the section Tasmannia is far more complicated than it seemed to be in 1943. This induced me to make a field study in New Guinea and to pay special attention to its morphology in the hope of finding additional criteria. On this basis the Australian species of Drimys could be better defined but the characters of the New Guinean taxa (Smith: 29 species) proved to be highly unstable. It was found that in the section Tasmannia the stomata are occluded by waxy plugs. The structure of the inflorescence is described and compared with that in section Drimys. In the flower two zones are distinguished. In the lower zone the floral appendages are arranged in (sub)opposite pairs, in the higher zone they are arranged according to the space available at the greatest possible distance from a focus of the floral apex which I termed ‘activity centre’; this centre often does not coincide with the ‘topographical apex’. The tilt of the floral apex appears to influence the arrangements as well. Shifting of determination factors in relation to the available locations on the floral apex is assumed to play an important role. As for the specific delimitation, the New Guinean material appeared to be the crux. The very heterogeneous material did not show correlated discontinuities in characters, which made it impossible to reach a division into species. The variability of the resulting very complex species Drimys piperita is described with the aid of a non-taxonomic subdivision, the units of wich are called ‘entities’. Field studies revealed that local populations are often very stable, but the morphological discontinuities between such local populations are obscured by material from other localities. It is supposed that vegetative propagation as observed in the field, combined with incidental cross-breeding, may be of importance in establishing such a situation. In Australia besides D. piperita four other species can be distinguished. The sections of Drimys are maintained, although Ehrendorfer c.s. and Smith proposed raising them to generic rank. In Pseudowintera the arrangement within the inflorescence is spiral, not decussate as described earlier, and a terminal flower is present. Minute hairs on bracts and young leaves, the mode of opening of the calyx cupule, and details of the lower leaf surface are used as a new set of specific characters. The fact that about twenty per cent of the specimens show varying combinations of specific characters is thought to be caused by hybridisation.