Models for the development and expression of symptoms in premenstrual syndrome (PMS) must integrate a variety of reported characteristics of the syndrome, such as the symptomatic profile, the menstrual cycle phase-specific timing of the symptoms, treatment response characteristics, and the absence of consistently observed biochemical or psychological markers. In this article, the authors review the available evidence supporting three proposed models for PMS. They also discuss a fourth model that may serve to explain and integrate many of the ostensibly conflicting observations of PMS and direct research toward the further characterization of this syndrome. These models are (1) biochemical or endocrine models; (2) symptom exaggeration or "special sensitivity" models; (3) premenstrual syndrome as a variant of affective disorder; and (4) premenstrual syndrome as a disorder of state regulation.