This paper suggests that a religion is not, as Freud proposed, a science-like thing, refutable by evidence. It is a socially constructed and maintained system of internal objects, analogous to those spoken of in psychoanalysis. Like analytic internal objects, religious objects have a heuristic function but no material existence. Unlike analytic objects, they are derived from a definite cultural tradition and are elaborated over time to meet the experience of practitioners. They may be understood to have a function of 'containing' the feelings, thoughts and fantasies arising in individual practitioners, and of making these experiences comprehensible. These objects then enable the believer to speak more truthfully of, and relate more fully to, the larger matrix within which the human world is situated. Finally, the suggestion is made that, as with analytic internal objects, religious objects may best be judged by their long-term effects on the lives of believers.