OBJECTIVE: To review the scientific evidence supporting an association between unemployment and adverse health outcomes and to assess the evidence on the basis of the epidemiologic criteria for causation. DATA SOURCES: MEDLINE was searched for all relevant articles with the use of the MeSH terms "unemployment," "employment," "job loss," "economy" and a range of mortality and morbidity outcomes. A secondary search was conducted for references from the primary search articles, review articles or published commentaries. Data and definitions of unemployment were drawn from Statistics Canada publications. STUDY SELECTION: Selection focused on articles published in the 1980s and 1990s. English-language reports of aggregate-level research (involving an entire population), such as time-series analyses, and studies of individual subjects, such as cross-sectional, case-control or cohort studies, were reviewed. In total, the authors reviewed 46 articles that described original studies. DATA EXTRACTION: Information was sought on the association (if any) between unemployment and health outcomes such as mortality rates, specific causes of death, incidence of physical and mental disorders and the use of health care services. Information was extracted on the nature of the association (positive or negative), measures of association (relative risk, odds ratio or standardized mortality ratio), and the direction of causation (whether unemployment caused ill health or vice versa). DATA SYNTHESIS: Most aggregate-level studies reported a positive association between national unemployment rates and rates of overall mortality and mortality due to cardiovascular disease and suicide. However, the relation between unemployment rates and motor-vehicle fatality rates may be inverse. Large, census-based cohort studies showed higher rates of overall mortality, death due to cardiovascular disease and suicide among unemployed men and women than among either employed people or the general population. Workers laid off because of factory closure have reported more symptoms and illnesses than employed people; some of these reports have been validated objectively. Unemployed people may be more likely than employed people to visit physicians, take medications or be admitted to general hospitals. A possible association between unemployment and rates of admission to psychiatric hospitals is complicated by other institutional and environmental factors. CONCLUSIONS: Evaluated on an epidemiologic basis, the evidence suggests a strong, positive association between unemployment and many adverse health outcomes. Whether unemployment causes these adverse outcomes is less straightforward, however, because there are likely many mediating and confounding factors, which may be social, economic or clinical. Many authors have suggested mechanisms of causation, but further research is needed to test these hypotheses.