Aim. The aim of the study was the adaptation and normalization of the Eating Disorder Inventory in the group of Polish schoolgirls and patients with eating disorders. Method: 297 healthy schoolgirls (14-22 y.o.) and 40 patients with anorexia nervosa and bulimia nervosa were examined using the EDI. Results. A Polish version of EDI revealed a high reliability in most of the subscales. Percentile norms were established. Comparison of subgroups of patients with restrictive and bulimic type of eating disorder revealed differences in most of the subscales. Conclusions. The Polish version of EDI is a useful research instrument. Differences between bulimic and restrictive groups of patients indicates that in severe cases of restrictive type of anorexia nervosa, self-evaluation using self-report questionnaires may underestimate the severity of symptoms.