Unemployment is known to be associated with poor mental health, but it is not clear how strongly unemployment leads to onset of diagnosed clinical depression (causation), or if depression raises the risks of becoming unemployed (health selection), or indeed if both pathways operate. We therefore investigate the direction of associations between clinical depression and unemploynent in a cross-cultural prospective cohort study. 10,059 consecutive general practice attendees (18-75 years) were recruited from six European countries and Chile between 2003 and 2004 and followed up at six, 12 and (in a subset) 24 months. The analysis sample was restricted to 3969 men and we men who were employed or unemployed and seeking employment and had data on depression measures. The outcomes were depressive episodes, assessed using the Depression Section of the Composite International Diagnostic Interview (CIDI) and self-reported employment status. Among 3969 men and women with complete data on depression and unemployment, 10% (n = 393) had depression symptoms and a further 6% (n = 221) had major depression at 12 months. 11% (n = 423) of the sample were unemployed by 6 months. Participants who became unemployed between baseline and 6 months compared to those employed at both times had an adjusted relative risk ratio for 12-month depression of 1.58 (95% Confidence Interval 0.76, 3.27). Participants with depression at baseline and 6 months compared to neither time had an odds ratio for 6-month unemployment of 1.58 (95% Confidence Interval 0.97, 2.58). We found evidence that causation and (to a lesser extent) health selection raise the prevalence of depression in the unemployed. Unemployed adults are at particular risk for onset of major clinical depression and should be offered extra services or screened. Given the trend for adults with depression to perhaps be at greater risk of subsequent unemployment, employees with depressive symptoms should also be supported at work as a precautionary principle. (C) 2011 Elsevier Ltd. All rights reserved.