Examination of data on patients admitted to psychiatric facilities in England in 1972-81, inclusive, with a main diagnosis of anorexia nervosa showed that first admissions for women rose significantly with time--age-period-cohort analysis showed that this was due entirely to an increase in the number of young women in the population rather than to an increase in risk of morbidity. There were significant increases with time in the readmission rates for both sexes. For the women, these were due to a complex interrelation between age, period, and cohort effects. This rising readmission rate may have contributed to the widely held impression that the incidence of anorexia nervosa has risen sharply.