The received wisdom on how activity affects energy expenditure is that the more activity is undertaken, the more calories will have been burned by the end of the day. Yet traditional huntergatherers, who lead physically hard lives, burn no more calories each day than western populations living in labour-saving environments. Indeed, there is now a wealth of data, both for humans and other animals, demonstrating that long-term lifestyle changes involving increases in exercise or other physical activities do not result in commensurate increases in daily energy expenditure (DEE). This is because humans and other animals exhibit a degree of 'energy compensation' at the organismal level, ameliorating some of the increases in DEE that would occur from the increased activity by decreasing the energy expended on other biological processes. And energy compensation can be sizable, reaching many hundreds of calories in humans. But the processes that are downregulated in the long-term to achieve energy compensation are far from clear, particularly in humans -we do not know how energy compensation is achieved. My review here of the literature on relevant exercise intervention studies, for both humans and other species, indicates conflict regarding the role that basal metabolic rate (BMR) or low level activity such as 'fidgeting' play, if any, particularly once changes in body composition are factored out. In situations where BMR and low-level activity are not major components of energy compensation, what then drives it? I discuss how changes in mitochondrial efficiency and changes in circadian fluctuations in BMR may contribute to our understanding of energy management. Currently unexplored, these mechanisms and others may provide important insights into the mystery of how energy compensation is achieved.