Brain glucose uptake is impaired in Alzheimer's disease (AD). A key question is whether cognitive decline can be delayed if this brain energy defect is at least partly corrected or bypassed early in the disease. The principal ketones (also called ketone bodies), β-hydroxybutyrate and acetoacetate, are the brain's main physiological alternative fuel to glucose. Three studies in mild-to-moderate AD have shown that, unlike with glucose, brain ketone uptake is not different from that in healthy age-matched controls. Published clinical trials demonstrate that increasing ketone availability to the brain via moderate nutritional ketosis has a modest beneficial effect on cognitive outcomes in mild-to-moderate AD and in mild cognitive impairment. Nutritional ketosis can be safely achieved by a high-fat ketogenic diet, by supplements providing 20–70 g/day of medium-chain triglycerides containing the eight- and ten-carbon fatty acids octanoate and decanoate, or by ketone esters. Given the acute dependence of the brain on its energy supply, it seems reasonable that the development of therapeutic strategies aimed at AD mandates consideration of how the underlying problem of deteriorating brain fuel supply can be corrected or delayed.