No scientific disclipine can be credited with much accuracy until it has succeeded in defining its principal terms. While recent contributions to the still youthful field of the psychology of religion have made the problems clearer, and at least partially solved some of them, no definition that has yet been proffered has met with wide acceptance. It is the purpose of this paper to suggest a working definition which, it is hoped, will prove to be reasonably satisfactory. It will be claimed that the definition adequately covers the established facts in regard to religion, that it furnishes a clear line of demarkation between religion and related fields, and that it is serviceable as a preliminary step toward the solution of the more difficult questions of the social significance and metaphysical validity of religion.2 An adequate definition of religion must satisfy two prerequisites. The first of these is that, to meet the requirements of logic, it must be exactly coextensive with the term defined.3 It must include all varieties of all religions in the past as well as the present, and all of the logically possible forms that religions may conceivably assume in the future, no matter whether we believe these forms to be desirable or undesirable, uplifting or degrading, true or false. It must be thoroughly impersonal and descriptive, not normative.