This paper reports the main findings of a large-scale study of subsidized psychoanalysis and long-term psychotherapy. More than 400 people in various phases, before, during and after subsidized psychoanalysis or long-term psychodynamic psychotherapy, were followed up for a period of three years with personal interviews, questionnaires and official statistics. Our analyses revealed progressive improvement the longer patients were in treatment--impressively strong among patients in psychoanalysis--on self-rating measures of symptom distress and morale. Improvement, however, was equally weak in both groups on a self-rating measure of social relations. Dosage factors (treatment duration and session frequency in combination) partly accounted for the outcome differences between those referred to psychoanalysis and those referred to long-term psychotherapy. Attitudes and ideals among therapists and analysts concerning the goals and means of psychotherapy were also associated with patient outcome, although in rather complex ways. A significant part of the outcome differences between patients in psychoanalysis and in psychotherapy could be explained by the adoption, in a large group of therapists, of orthodox psychoanalytic attitudes that seemed to be counterproductive in the practice of psychotherapy but not in psychoanalysis. It is suggested that this effect may be a negative transfer of the psychoanalytic stance into psychotherapeutic practice and that this may be especially pronounced when the attitudes are not backed up by psychoanalytic training.