The author examines two positions of the patient in relation to the work of mourning. In the first position, mourning is accepted within the complicated interplay of hysterical and narcissistic identifications, the dynamics of the transference, and the regressions attributable to the analytic situation. In the second position, the work of mourning is somehow eluded without the patient becoming psychotic or developing a narcissistic neurosis. This is possible because of the establishment of what the author calls 'nostalgic relationship to the object'. The author gives clinical examples from which he derived his hypothesis. He gives nostalgia the status of a condition having a specific type of object relations, and goes to considerable lengths to justify the hypothesis theoretically.