The purpose of this paper is to describe two particular ideas, which we shall call "anxiety conservation" and "partial irreversibility," within a general theory of anxiety and avoidance learning. In doing so, more familiar postulates appear to us to generate some new and interesting theorems about behavior, some of which seem to correspond to established facts. We shall not concern ourselves at this time with an exhaustive review of empirical data. Rather, we shall direct our attention to a theoretical argument and shall illustrate specific points with observations drawn from the following fields: avoidance learning, psychotherapy, physiological psychology, and psychosomatic medicine.Although we shall make no attempt here to validate thoroughly the logical deductions from our theoretical notions, 1 We wish to thank our colleagues for many instructive and constructive criticisms of the preliminary drafts of this manuscript. We are especially indebted to G.