The concept that thermogenesis in brown adipose tissue can play a role as an energy buffer has developed during the last 5 years. The history of this development is reviewed. Control of brown adipose tissue thermogenesis resides in regions of the brain, located primarily but not exclusively in the hypothalamus, that control the activity of the sympathetic nervous system in response to diet and to environmental temperature. Brown adipose tissue mitochondria are uniquely specialized for thermogenesis, possessing a specific proton leakage mechanism that is regulated by the concentration of fatty acids in the cells of the brown adipose tissue. The level of fatty acids is in turn controlled by the lipolytic action of noradrenaline on the tissue. Sympathetic stimulation also exerts a trophic influence on brown adipose tissue. Effective thermogenesis in brown adipose tissue is associated with leanness and decreased metabolic efficiency, as in the rat rendered hyperphagic and hypermetabolic, by either cold acclimation or cafeteria feeding. Conversely, food restriction is associated with suppressed thermogenesis in brown adipose tissue and increased metabolic efficiency. Defective brown adipose tissue thermogenesis is associated with obesity in a number of different types of obese animals. In three of these (the genetically obese fa/fa Zucker rat, the mouse with hypothalamic damage induced by gold thioglucose, the rat with a surgically induced hypothalamic lesion), diet-induced thermogenesis is defective in brown adipose tissue, but cold-induced thermogenesis is normal. In another type of obese animal, the genetically obese (ob/ob) mouse, control of brown adipose tissue is defective. Studies of this control are complicated by the frequency of torpor in the fed state. The unusual cold sensitivity of the ob/ob mouse is probably due to inappropriate entry into torpor and coincident suppression of brown adipose tissue thermogenesis. Whether the knowledge that defective thermogenesis in brown adipose tissue contributes to obesity in laboratory animals can be applied to humans is presently under investigation. Current research and problems in this area are discussed.