The available treatment literature supports a role for medication management in the treatment of both AN and BN. The data on BN are clearer and suggest that antidepressant therapy - fluoxetine being the drug most widely studied - is superior to treatment with placebo but less effective than CBT alone, with one such study suggesting that the combination may provide optimal treatment. Specific recommendations as to when to add or not add antidepressants to CBT have been made, although the rules suggested here have yet to be empirically tested. Although the data on AN are much more limited, information available suggests a lack of efficacy for SSRIs in patients with AN at low weight and considerable use for SSRIs when used in combination with psychotherapy for patients with AN following weight recovery. Where do we go from here? Several pressing issues require careful study. First, in the case of patients with AN, can other agents, in particular the new atypical antipsychotics, be useful in treating patients when they are at low weight? In terms of relapse prevention, can the available findings indicating a role for antidepressants in relapse prevention be replicated, and, if so, can predictor variables that are associated with antidepressant response be identified? In the case of BN, clinicians need to know more about the best possible way to sequence interventions. It has been proposed to add medication to CBT early in treatment if the response to CBT alone is thought to be inadequate. However, other models should be considered, such as stepped-care models in which self-help manuals are used in conjunction with medications. The advantage of these interventions is they could be made more widely available than CBT, which requires a specialist's care. Also, several other new agents, such as sibutramine, which is a drug with serotonin and norepinephrine reuptake inhibition effects, should be tested empirically in subjects with BN, given their pharmacologic profiles.