OBJECTIVES In Japan, estimated numbers of patients who received medical treatment with major depressive disorder have recently increased and are reported almost 708,000 patients according to the results of the Patient Survey 2011. In addition, the potential magnitude of disease burden associated with undiagnosed depression is suggested by the low mental health service utilization in Japan, which is even lower than many western countries. The aim of this study is to investigate whether depression severity was associated with work impairments, regardless of diagnosis. METHODS Employed respondents (n=17,820) of an internet-based cross-sectional survey (the 2011 National Health and Wellness Survey fielded in Japan) were categorized according to depression diagnosis (diagnosed or undiagnosed) and severity (Patient Health Questionnaire-9 score=10+ or <10). Work Productivity and Activity Impairment questionnaire-based measures were predicted by diagnosis, severity, and their interaction, adjusting for covariates. RESULTS Among employed respondents (n=17,820) in the Japan NHWS, 3.8% were diagnosed with depression within 12 months (n=678). Among those with diagnosis, 51.0% (n = 346) had PHQ-9 scores of 10 and above, while among those without a diagnosis, 7.8% (n = 1,336) had scores of 10+. In other words, 7.5% among all employed respondents (1,336/17,820) self-reported that they were undiagnosed as depressed but had PHQ-9=10+. Among the undiagnosed, high severity respondents had greater overall work impairment (33.3% vs. 14.8%), absenteeism (5.9% vs. 2.2%), presenteeism (30.4% vs. 13.4%) than low severity respondents. Significant interactions between diagnosis and severity indicated greater impairments among undiagnosed vs. diagnosed respondents (except on absenteeism). CONCLUSIONS Depression severity was associated with work productivity loss, even among the undiagnosed, suggesting a need for early detection, referral, and treatment of depression in the workplace.