Aims Metabolic Syndrome (MetS) is a major health epidemic of Western countries and patients with schizophrenia is a particularly vulnerable population due to lifestyle, mental illness and treatment factors. However, we lack prospective data to guide prevention. The aim of our study is then to determine MetS incidence and predictors in schizophrenia. Method Participants were recruited in 10 expert centers at a national level and followed-up for 3 years. MetS was defined according to the International Diabetes Federation criteria. Inverse probability weighting methods were used to correct for attrition bias. Results Among the 512 participants followed-up for 3 years, 77.9% had at least one metabolic disturbance. 27.5% were identified with MetS at baseline and excluded from the analyses. Among the rest of participants (N = 371, mean aged 31.2 (SD = 9.1) years, with mean illness duration of 10.0 (SD = 7.6) years and 273 (73.6%) men), MetS incidence was 20.8% at 3 years and raised to 23.6% in tobacco smokers, 29.4% in participants receiving antidepressant prescription at baseline and 42.0% for those with 2 disturbed metabolic disturbances at baseline. Our multivariate analyses confirmed tobacco smoking and antidepressant consumption as independent predictors of MetS onset (adjusted odds ratios (aOR) = 3.82 [1.27–11.45], p = 0.016, and aOR = 3.50 [1.26–9.70], p = 0.0158). Antidepressant prescription predicted more specifically increased lipid disturbances and paroxetine was associated with the highest risk of MetS onset. Conclusion These results are an alarm call to prioritize MetS prevention and research in schizophrenia. We have listed interventions that should be actively promoted in clinical practice.< Réduire