An important finding of the Collaborative Depression Study that patients and astute clinicians have always known (but not always paid attention to) is that the standard definitions of response (50% improvement) and remission (depression or mania rating scale scores that are about two-thirds better but not zero), which are the primary outcome measures of the clinical trials on which we all base treatment choices, are not satisfactory outcomes. Most of us would not be satisfied with feeling better but not well, and any residual symptoms greatly increase the risk of major relapse and recurrence. Even mild, subsyndromal symptoms are associated with significant levels of impairment, which is an independent predictor of relapse. Indeed, the observation that “many individuals have residual subsyndromal symptoms and disabling psychosocial impairment when recovered from major episodes of depression for at least 2 months” (p. 169) indicates that the true meaning of “recovery” does not apply to many patients. The investigators point out that residual depressive symptoms deserve as much aggressive attention as the less common residual manic and hypomanic symptoms in bipolar disorder, and they suggest that antidepressants may not worsen the course of all cases of bipolar disorder. However, they are not yet able to reassure us about which bipolar disorder patients can and which cannot tolerate ongoing treatment with antidepressants. Despite the widely acknowledged need to treat mood disorders early and aggressively, to combine pharmacologic and psychological therapies, and to reduce the risk of relapse by continuing whatever treatment is effective acutely, mood disorders remain undertreated.