Here we revisit and reinterpret the original study in which the so-called 'Maudsley (London) model' of family therapy was compared with individual therapy for anorexia nervosa. Family therapy was more effective in adolescents with a short duration of illness. However, this is only part of the story. A later study describing the 5-year outcome contains important information. Those adolescents randomised to family therapy achieved a better outcome 5 years later. Moreover, the group with an onset in adolescence but who had been ill for over 3 years had a poor response to both family and individual therapy, suggesting that unless effective treatment is given within the first 3 years of illness onset, the outcome is poor. We examine other evidence supporting this conclusion and consider the developmental and neurobiological factors that can account for this.