BACKGROUND: Few studies have systematically compared the relationship between lost work productivity (work impairment) and mental disorders using population surveys. AIMS: (1) To identify the importance of individual mental disorders and disorder co-occurrences (comorbidity) as predictors of two measures of work impairment over the past month - work loss (number of days unable to perform usual activities) and work cutback (number of days where usual activities were restricted); (2) to examine whether different types of disorder have a greater impact on work impairment in some occupations than others; (3) to determine whether work impairment in those with a disorder is related to treatment seeking. METHOD: Data were based on full-time workers identified by the Australian National Survey of Mental Health and Well-Being, a household survey of mental disorders modeled on the US National Comorbidity Survey. Diagnoses were of one-month DSM-IV affective, anxiety and substance-related disorders. Screening instruments generated likely cases of ICD-10 personality disorders. The association of disorder types and their co-occurrences with work impairment was examined using multivariate linear regression. Odds ratios determined the significance of mental disorder prevalence across occupations, and planned contrasts were used to test for differences in work impairment across occupations within disorder types. The relationship between work impairment and treatment seeking was determined for each broad diagnostic group with t-tests. RESULTS: Depression, generalized anxiety disorder and personality disorders were predictive of work impairment after controlling for impairment due to physical disorders. Among pure and comorbid disorders, affective and comorbid anxiety-affective disorders respectively were associated with the greatest amount of work impairment. For all disorders, stronger associations were obtained for work cutback than for work loss. No relationship was found between type of occupation and the impact of different types of disorder on work impairment. Only 15% of people with any mental disorder had sought help in the past month. For any mental disorder, significantly greater work loss and work cutback was associated with treatment seeking, but comparisons within specific disorder types were not significant. DISCUSSION: A substantial amount of lost productivity due to mental disorders comes from within the full-time working population. The greater impact of mental disorders on work cutback compared to work loss suggests that work cutback provides a more sensitive measure of work impairment in those with mental disorders. Work impairment was based on self-report only. While there is evidence for the reliability of self-assessed work loss days, no reliability or validity studies have been conducted for work cutback days. The low rates of treatment seeking are a major health issue for the workforce, particularly for affective and anxiety disorders, which are important predictors of lost productivity. IMPLICATIONS FOR HEALTH POLICIES AND FURTHER RESEARCH: Future research should investigate the validity of work cutback, given its importance as a measure of lost productivity in people with mental disorders. Employers need to be aware of the extent to which mental disorders affect their employees so that effective work place interventions can take place. Treatment should be targeted at people with affective and anxiety disorders, particularly where they co-occur.