Twenty years after the start of the Boston outcome study of psychoanalysis, the author reviews the findings of this investigation. Twenty-two patients had been seen in four or five times a week supervised psychoanalysis by candidates at the Boston Psychoanalytic Institute. The underlying assumptions, on which this project and other outcomes studies were based, are reconsidered in light of contemporary understanding of the analytic process and the analytic relationship. The impact of the patient-analyst match on the outcome of treatment may help to explain why some seemingly suitable patients have unsuccessful outcomes and some seemingly unsuitable patients have favorable outcomes.