Schema therapy has proven to be an effective treatment for patients with borderline personality disorder. However, little is known of its merits in other psychiatric (personality) disorders. Objective This study investigated whether schema therapy in a group setting (group schema cognitive–behavioural therapy [SCBT‐g]) was associated with changes in symptom and schema and mode severity. Furthermore, the aim was to search for baseline predictors and possible mediators of treatment outcome. Design and method Sixty‐three heterogeneous psychiatric outpatients who attended the SCBT‐g were included as participants. In this naturalistic pre‐treatment and post‐treatment design, data were available on the Symptom Checklist 90, the Schema Questionnaire and the Young–Atkinson Mode Inventory. Results All outcome measurements showed changes with moderate to high effect sizes, with 53.2% of the patients showing a significant reduction in severity of psychiatric symptoms and schemas and modes. Higher pre‐treatment levels of the schema domain Other Directedness predicted greater symptom reduction. Pre‐treatment to mid‐treatment changes in schema severity predicted subsequent symptom improvement, but change in symptoms and schemas proved to be strongly correlated. Conclusions In this naturalistic study, SCBT‐g was associated with reduced symptom and schema and mode severity in more than half of the psychiatric outpatients. Furthermore, the results suggest that changes in schemas and symptomatology mutually reinforce each other. Copyright © 2012 John Wiley & Sons, Ltd. Key Practitioner Message Over 50% of ambulatory patients show clinical improvement after treatment in a short‐term schema therapy group. Other Directedness seems to be a predictor of schema group therapy success. More randomized controlled trial studies and prediction and mediation studies on (short‐term) schema group therapy are sorely needed.