The impact of work on the risk of future psychiatric disorder has been examined in few longitudinal studies. This was examined prospectively in a large epidemiological study of civil servants. In the Whitehall II study, a longitudinal, prospective cohort study of 6895 male and 3413 female London based civil servants, work characteristics measured at baseline (phase 1: 1985-8) and first follow up (phase 2: 1989) were used to predict psychiatric disorder measured by a 30 item general health questionnaire (GHQ) at phase 2 and phase 3 follow up (phase 3: 1991-3). Work characteristics and GHQ were measured at all three phases. Low social support at work and low decision authority, high job demands and effort-reward imbalance were associated with increased risk of psychiatric disorder as assessed by the GHQ at follow up adjusting for age, employment grade, and baseline GHQ score. Social support and control at work protect mental health while high job demands and effort-reward imbalance are risk factors for future psychiatric disorder. Intervention at the level of work design, organisation, and management might have positive effects on mental health in working populations.