The scaling of metabolic rate with body mass has long been a controversial topic. Some workers have claimed that the slope of log-log metabolic scaling relationships typically obeys a universal 3/4-power law resulting from the geometry of resource-transport networks. Others have attempted to explain the broad diversity of metabolic scaling relationships. Although several potentially useful models have been proposed, at present none successfully predicts the entire range of scaling relationships seen among both physiological states and taxonomic groups of animals and plants. Here I argue that our understanding may be aided by three shifts in focus: from explaining average tendencies to explaining variation between extreme boundary limits, from explaining the slope and elevation (metabolic level) of scaling relationships separately to showing how and why they are interrelated, and from focusing primarily on internal factors (e.g. body design) to a more balanced consideration of both internal and external (ecological) factors. By incorporating all of these shifts in focus, the recently proposed metabolic-level boundaries hypothesis appears to provide a useful way of explaining both taxonomic and physiological variation in metabolic scaling relationships. This hypothesis correctly predicts that the scaling slope should vary mostly between 2/3 and 1 and that it should be related to metabolic (activity) level according to an approximately U-shaped function. It also implies that the scaling of other energy-dependent biological processes should be related to the metabolic level of the organisms being examined. Some data are presented that support this implication, but further research is needed.