INTRODUCTION Psychiatric epidemiologists were among the first to use the term “social epidemiology” (1), and the role of the social environment in the etiology and course of major mental disorders continues to be investigated (2–5). A number of reviews published in the late 1990s documented the associations between socioeconomic position (SEP) and specific mental disorders (6–9); in 2003, a comprehensive meta-analysis of the research on SEP and depression (10) concluded that both prevalence and incidence studies show that persons of low SEP (i.e., low educational and low income levels) are at a higher risk of depression. Here, we examine innovative developments in the study of the associations between SEP and major mental disorders. We use the term “socioeconomic position” for pragmatic and conceptual reasons: 1) it allows us to follow the convention in the first textbook of social epidemiology (i.e., Berkman and Kawachi’s Social Epidemiology (8)), and 2) socioeconomic “position” is neutral with respect to the relational/ordinal distinction that sets social class apart from socioeconomic status. Thus, the term “socioeconomic position” encompasses both social class (referring to social relations of ownership and control over productive assets) and socioeconomic status (referring to the ordering of persons along a continuum of some valued socioeconomic attribute such as income or education). We focus on the life-course approach (11) taken in studies of the selection-causation issue, using data on ethnic stratification, immigration and schizophrenia, and the long-term impact of early life-course exposures such as fetal stress and childhood poverty. We also highlight new developments in social class concepts and measures that have led to new findings on the effects of SEP, the relative contributions of neomaterial and psychosocial pathways, evidence on multilevel associations between geographic area SEP and mental disorders, and gender-specific hypotheses. Contrary to the idea that the research on the relation between SEP and mental disorders has been exhausted, these recent innovations are generating promising hypotheses to be tested in future research in upcoming years. Using multidisciplinary (Current Contents; Thomson ISI, Philadelphia, Pennsylvania) and biomedical (PubMed; National Library of Medicine, Bethesda, Maryland) databases, our review is based primarily on English-language cross-sectional and longitudinal studies published between 1999 and August 2003 that included “social class” or “socioeconomic status” and selected major psychiatric disorders (i.e., schizophrenia, major depression, and anxiety disorders). The period covered was chosen to minimize overlap with previous reviews (6, 7, 9, 12). Disorders were selected on the basis of their prevalence in the population (depression and anxiety) or their centrality to the literature on SEP (schizophrenia, bipolar disorder, and depression). Evidence on the relation of anxiety disorders to SEP is more limited because the diagnosis has been subject to more fluctuation than that for depression, bipolar disorder, or schizophrenia. For example, the current diagnostic concepts of panic disorder, phobic disorder, and generalized anxiety disorder are new with the 1980 revision of the diagnostic manual of the American Psychiatric Association (13). As a result, few longitudinal studies are available. Prevalence data from the Epidemiologic Catchment Area Study (14–16) consistently suggest that lower socioeconomic status groups have a higher prevalence of panic, all types of phobias, and generalized anxiety disorder. The evidence is less conclusive for obsessive-compulsive disorder. These general findings about socioeconomic status and prevalence were confirmed in the National Comorbidity Study for generalized anxiety disorder (17), panic (18), and agoraphobia, simple phobia, and social phobia (19). Incidence data from the Epidemiologic Catchment Area Study show that low occupational prestige is a risk factor for panic attacks and panic disorder (20) and that low educational level is a risk factor for agoraphobia (21), social phobia (22), and obsessive-compulsive disorder (23). To our knowledge, there are no population-based incidence data for generalized anxiety disorder. Since we did not find much innovation in the literature on SEP and anxiety disorders or bipolar disorder, our review focuses on schizophrenia and depression. Because of the breadth of the SEP literature, we have omitted associated work on community integration and the role of access to and utilization of health services in explaining the relation between SEP and the course of major psychiatric disorders (24).