Stephen A. Mitchell's theoretical writings are examined, with particular emphasis on his differences with drive theory, the consequences of his recommendation for greater involvement by the analyst, his (naive) trust in psychoanalytic self-correction, and the problems that follow from his belief that empirical data have little to contribute to psychoanalysis.Stephen A. Mitchell died unexpectedly on December 21, 2000, at the age of 54. It was a tragic loss because he was already one of the most influential and articulate leaders in psychoanalysis and he had a great deal more to contribute. His book with Greenberg Object Relations in Psychoanalytic Theory (Greenberg & Mitchell, 1983) sold 44,000 copies in the English edition alone (Spezzano, 2001, p. 10) and was instrumental in catapulting object relations theories into a leading place in American psychoanalysis. It was followed by Relational Concepts in Psychoanalysis (Mitchell, 1988), Hope and Dread in Psychoanalysis (Mitchell, 1993), and Influence and Autonomy in Psychoanalysis (Mitchell, 1997).His obituary notices described a man of enormous enthusiasm, discipline, and intelligence, whose caring and warmth attracted huge numbers of students and followers, making him "the most sought after teacher at NYU and at White, a mentor beyond parallel" (Ghent, 2001, p. 7). A founder of the relational school in psychoanalysis, he was a thoughtful (and persistent) critic of the psychoanalytic drive model. In contrast to a good deal of psychoanalytic writing, so often turgid, tedious, and impenetrable, Mitchell's prose is so well written it looks effortless. Friends and critics alike would have to agree that few have had such powerful influence on contemporary psychoanalysis.I am indebted to a number of colleagues for their counsel in the preparation of this essay. Drs.