The phylogeny of the pandas has recently been actively debated (O'Brien et al. 1985, Kurten 1985, Tagle et al. 1986, Van Valen 1986). Mayr (1986), in a review, argued that the evidence in favour of assigning the giant panda (Ailuropoda melanoleuca) to the family Ursidae (bears) is overwhelming, even in the light of ostensibly conflicting data from haemoglobin protein sequencing (Tagle et al. 1986). Although Mayr's (1986) arguments were criticized for tautological reasoning by Warr (1986), we agree with Mayr's thesis and would like to offer additional and independent support for assigning pandas to the ursid lineage. Giant pandas have short gestation and neonates are anomalously tiny at birth relative to the size of their mother (Kleiman 1983, Schaller et al. 1985); a pattern seen in all extant ursids but not known in any other eutherian mammals (Frazer and Huggett 1974, McKeown et al. 1976). After parturition, female pandas carry their tiny neonates with them as they forage for bamboo (Schaller et al. 1985); in comparison, gestation and early postpartum development in Holarctic bears takes place while the mother is in a state of dormancy in a winter den, without food or water (Wimsatt 1963, Nelson 1980). Throughout the ursid denning period, maternal fat stores (which are usually considerable) must supply the metabolic energy requirements of mother and offspring. We have suggested that bears have such a short gestation and low birth weight because of constraints on the ability of mammalian foetuses to catabolize free fatty acids (FFA); they require carbohydrates (Ramsay and Dunbrack 1986). Thus, foetuses cannot use maternal fat stores as an energy source; but once born, the young can, in contrast, readily oxidize FFA and thus make use of the mother's stored fat in the form of milk. Early birth therefore means early access to an easily transferred source of energy. Pandas have access to carbohydrates year round (Schaller et al. 1985), so their production of notably small neonates does not appear to be a response to extended fasting in a den; but small young could be explained if the suite of physiological and behavioural adaptations associated with the production of small neonates were established before splitting of the panda and ursid lines. Since the denning behaviour accompanying extended fasting appears to have developed early in the lineage leading to modern bears (Hunt et al. 1983), probably in response to seasonally fluctuating food resources, we suggest that the remarkably bear-like reproductive pattern of the giant panda, in the absence of extended fasting and denning, is due to phylogenetic constraints and, thus, argues in favour of an ursid ancestry.