There is currently no evidence-based definition of severe and enduring anorexia nervosa (SE-AN) with which to reliably inform clinical practice and research. Indeed, data on the effect of AN severity and duration on treatment outcome are inconsistent. A large group of patients with SE-AN are repeatedly unsuccessfully managed with the available eating disorders treatments and have no access to adequate treatment for their illness. Cognitive behavioral therapy (CBT) adapted for SE-AN has been designed to enhance quality of life and reduce harm rather than promoting weight gain in such patients, and has had some success. However, a percentage of patients with SE-AN achieves remission, or at least returns to a normal weight range, with available evidence-based treatments for eating disorders, such as enhanced CBT (CBT-E). It would therefore be worth conducting a large-scale randomized controlled trial comparing CBT adapted for SE-AN with CBT-E to assess their relative acceptability; efficacy, including their effect on quality of life and medical stability; cost-effectiveness; and the treatment response moderators that might allow better matching of patients with SE-AN to a treatment oriented either to harm reduction or to change.