BACKGROUND: Individual Placement and Support (IPS) has been demonstrated to increase return to open employment significantly in individuals with mental health problems in the USA. Previous experience (e.g. with assertive community treatment) has demonstrated the sensitivity of complex community mental health interventions to local social and healthcare cultures. Europe has conditions of generally greater employment security than the USA, and varying (generally higher) unemployment rates and welfare benefits. Evidence of the effectiveness of IPS in these conditions, and its potential variation across them, would guide local policy and provide possible insights into its mechanism. METHODS: We conducted a randomized controlled trial of IPS versus high-quality train-and-place vocational rehabilitation in six European centres with very different labour market and health and social care conditions. A sample of 312 individuals with psychotic illness was randomly allocated (50 per site). Inclusion criteria were a minimum of two years illness duration, with at least one year of continuous unemployment and six months contact with their current mental health services. Follow-up was 18 months. The primary outcome was any open employment, and secondary outcomes included time to employment, duration of employment and hospital admission. FINDINGS: IPS was more effective than the vocational services for all vocational outcomes. 85 IPS patients (54.5%) worked for at least one day compared to 43 vocational service patients (27.6%). They were significantly less likely to have been rehospitalized. Local unemployment rates explained a significant amount of the variation in IPS effectiveness and both national economic growth and welfare systems influenced overall employment rates in both services. CONCLUSIONS: IPS doubles the access to work of people with psychotic illnesses, without any evidence of increased relapse. Its effectiveness is not independent of external circumstances, particularly local unemployment rates.