The scientifically novel, but evolutionarily ancient, so-called uncoupling proteins 2 and 3 (UCP2, UCP3) are structually similar to the archetypical uncoupling protein UCP1. A series of suggestions have been forwarded for their physiological function. We discuss systematically here the pros and cons for these suggestions. We conclude that the novel UCPs do not seem to be physiologically relevant uncoupling proteins; the uncoupling property was apparently a late introduction into the subfamily through the evolution of UCP1. Physiological functions ascribed to UCP2 and UCP3 based on their purported uncoupling property may have to be revised (i.e. any type of thermogenesis, including protection against obesity, protection against the formation of reactive oxygen species and thermogenic involvement in the fever response). The presence of a mixed genetic background in most published studies of UCP2 or UCP3 gene-ablated mice also means that data concerning marked differences in diabetes propensity, infection sensitivity and production of reactive oxygen species may require confirmation in backcrossed mice. The increased expression of UCP2 and UCP3 under conditions of increased fatty acid metabolism implies an as yet undefined role in lipid metabolism. Thus, the novel UCPs should probably be considered as mitochondrial carriers, and the challenge now is to identify the transported molecule. Experimental Physiology (2003) 88.1, 65-84.