Aim of the study was to evaluate the clinical characteristics and the treatment outcome of underweight patients with eating disorder (ED) not otherwise specified without amenorrhea (EDNOS-WA), compared with classical anorexia nervosa (AN) cases. Seventy-three consecutive female patients (57 AN, and 16 EDNOS-WA) were evaluated before and after a 20-week cognitive behaviour inpatient treatment (CBT-I). Assessment included anthropometry, the Eating Disorder Examination (EDE), the Beck Depression Inventory (BDI), the State-Trait Anxiety Inventory (STAI), and the Temperament and Character Inventory (TCI). At logistic regression analysis, amenorrhea was only predicted by baseline BMI and intense exercise, not by psychopathological variables. Response to CBT-I was good and similar between groups, without differences in the dropout rate or time-to-dropout. Our data lend support to the hypothesis that the criterion "amenorrhea" is of no clinical utility in the diagnosis and treatment of AN and could be removed in the forthcoming DSM-V proposal.