Animals at rest and during exercise display rates of aerobic metabolism, VO2, that represent mainly the sum of mitochondrial respiration rates in various organs. The relative contributions of these organs change with physiological state such that internal organs such as liver, kidney and brain account for most of the whole-body VO2 at rest, while locomotory muscles account for >90% of the maximum rate, VO2max, during maximal aerobic exercise. Mechanisms that regulate VO2 are complex and the relative importance of each step in a series, estimated by metabolic control analysis, depends upon the level of biological organization under consideration as well as physiological state. Despite this complexity, prominent single-cause models propose that metabolic rates are supply-limited and that the scaling of supply systems provides a sufficient explanation for the allometric scaling of metabolism. We argue that some assumptions, as well as current interpretations of the meaning (or consequences) of these constraints are flawed, i.e., elephants do not have lower mass-specific basal or maximal rates of aerobic metabolism because their mitochondria are more supply-limited than those of shrews. Animals do not violate the laws of physics, and the allometric scaling of supply systems would be expected, to some extent, to be matched by capacities for (and rates of) energy expenditure. But life is not so simple. Animals are so diverse that to do justice to metabolic scaling, it is also necessary to consider the scaling of energy expenditure. It is by doing so that models of metabolic scaling can be consistent with current paradigms in metabolic regulation and accommodate the range of inter- and intraspecific exponents found in nature. The "allometric cascade," a first attempt at such an accounting, was a source of great satisfaction to Peter Hochachka. It was the last door that he helped open to comparative physiologists before he said goodbye.