OBJECTIVE: Energy restriction reduces the incidence of malignant tumors in experimental animals, but evidence for a similar effect in humans is lacking. To test the hypothesis in humans, we investigated cancer incidence among patients with anorexia nervosa, who have had an extremely low intake of calories for prolonged periods of their lives. METHODS: Patients with anorexia nervosa (2151 women and 186 men) during 1970-1993 were identified in the population-based Danish Psychiatric Case Register and the National Registry of Patients. The cohort was linked to the Danish Cancer Registry, and cancer incidence among cohort members was compared with that of the general population. RESULTS: The overall cancer incidence among women with anorexia nervosa was reduced by a factor of 0.80 (95% confidence interval 0.52-1.18) below that of the general population on the basis of 25 observed and 31.4 expected cases. Among men, two cases of cancer were observed, both confined to the brain, whereas 0.2 cases were expected. CONCLUSIONS: The finding of a slight reduction in cancer risk among women with anorexia nervosa may support the theory that a low-energy diet may decrease tumor development in humans. However, longer follow-up and control for confounding factors are needed to obtain more convincing evidence.