The existence of two sexes across most of the living world is strange. While sex and recombination is genetically beneficial, there is no obvious need for two sexes. All that is required is the fusion of two gametes, which could in principle be identical. The frequency of mating could potentially be doubled if all individuals were the same sex, which would offer a large selective advantage in widely dispersed populations. Yet even when gametes are identical, as in many algae and fungi, each gamete can only fuse with the opposite mating type, to the exclusion of 50% of the population. Many studies suggest that the unifying explanation is the inheritance of organelles, notably mitochondria. The deepest biological distinction between the sexes is that one sex (usually female) passes on mitochondria to the next generation, whereas the other sex (usually male) does not: male mitochondria are destroyed. In this talk I will put forward a new bioenergetic hypothesis for the evolution of two sexes. To achieve optimal energy coupling (the coupling of cell respiration to the synthesis of ATP by way of a proton gradient) mitochondrial genes need to be selected to function properly against the nuclear background of the new individual. This functional selection is optimised by the uniparental inheritance of mitochondria. In animals, selection for optimal energy coupling sets up an evolutionary tradeoff, in which the cost of high physical (aerobic) fitness and long lifespan is low fecundity and poor adaptability to different environments.