During 1972-74 nine populations of schoolgirls were surveyed in respect of the occurrence of anorexia nervosa. As well as the current school year, previous years (going back variously 3-6 years) were also carefully studied within seven of the schools. Anorexia nervosa was only identified as present or having been present if the diagnosis was unequivocal: for instance, seemingly evident and severe cases which could not be traced were excluded. The condition in severe form was found to be relatively common in the independent sector of education, probably implying a social class factor. In all such schools, the prevalence was one severe case in approximately every 200 girls. In those aged 16 and over it amounted to one severe case in about every 100 girls. Overall, these schools encountered one 'new' severe case in every 250 pupils aged 16 years and over. The condition is likely to be even more common than this study allows because of the stringent limits set here on inclusion of cases and the age-bands studied. It is concluded that severe anorexia nervosa is a common condition and is probably getting more common.