This study explored the association between psychosocial work characteristics and incidence of depression as indicated in three complementary models (the Job Strain Model, the Team Climate Model, and the Procedural Justice Model). Participants were 4815 Finnish hospital personnel (4278 women and 537 men) free from diagnosed depression at entry into the study. A baseline survey in 1998 measured psychosocial work characteristics, health-related behaviours, psychological distress, and doctor-diagnosed depression. The factor analysis of pooled questionnaire items on psychosocial work characteristics supported a five-factor solution with the following distinct dimensions: team climate, relational justice, procedural justice, job control, and job demands. Items in these dimensions were used as scales and job strain was modelled as a combination of job demands and job control. A follow-up survey in 2000 identified 225 incident cases of depression. After adjustment for age, sex and income, poor team climate, low procedural justice, and low relational justice were associated with a higher risk of new depression, the odds ratios (ORs) 1.58 (95% confidence interval (CI) 1.11-2.24), 1.45 (95% CI 1.03-2.04), and 1.39 (95% CI 1.00-1.96), respectively. After additional adjustment for lifestyle factors and exclusion of those with psychological distress at baseline, there was still an association between poor team climate and risk of depression (ORs 1.55 and 1.75, respectively). Job control, work demands, and job strain did not predict the 2-year incidence of depression, and the effects of all psychosocial work characteristics were attenuated when entered simultaneously in the model. In conclusion, work unit social factors seem to be predictive of subsequent doctor-diagnosed depression, but other aspects of psychosocial work environment may also be important.