In a consensus-building process a group of experts from 19 European countries (COST Action B6) adapted the terms partial and full remission, relapse, recovery, and recurrence according to principles described by Frank et al. for depression. The empirical validity of the operationalizations was illustrated by longitudinal data on the post treatment course of 233 anorectic and 422 bulimic patients (diagnosed according to DSM-IIIR) from the German Project TR-EAT. These data were collected 2.5 years after admission using the Longitudinal Interval Follow-up Evaluation (LIFE) and statistically explored by survival-analysis. It was demonstrated that these consensus definitions measure what they intend to measure. They open a longitudinal perspective in that one can learn not only whether, but also when and with what probability patients change for the better or worse. Data suggest that persistence of symptom improvement has different implications for anorexia and bulimia nervosa. For example, relapse prevention would be most beneficial for bulimic patients for about 4 months after key symptoms remit, while this would be of less importance for anorexic patients. It is discussed whether and how this longitudinal approach can contribute to an empirically based rationale for targeted and individualized treatment. © 2002 Wiley Periodicals, Inc. J Clin Psychol 58: 833–846, 2002