BackgroundAims were to assess the efficacy of metacognitive training (MCT) in people with a recent onset of psychosis in terms of symptoms as a primary outcome and metacognitive variables as a secondary outcome.MethodA multicenter, randomized, controlled clinical trial was performed. A total of 126 patients were randomized to an MCT or a psycho-educational intervention with cognitive-behavioral elements. The sample was composed of people with a recent onset of psychosis, recruited from nine public centers in Spain. The treatment consisted of eight weekly sessions for both groups. Patients were assessed at three time-points: baseline, post-treatment, and at 6 months follow-up. The evaluator was blinded to the condition of the patient. Symptoms were assessed with the PANSS and metacognition was assessed with a battery of questionnaires of cognitive biases and social cognition.ResultsBoth MCT and psycho-educational groups had improved symptoms post-treatment and at follow-up, with greater improvements in the MCT group. The MCT group was superior to the psycho-educational group on the Beck Cognitive Insight Scale (BCIS) total (p= 0.026) and self-certainty (p= 0.035) and dependence self-subscale of irrational beliefs, comparing baseline and post-treatment. Moreover, comparing baseline and follow-up, the MCT group was better than the psycho-educational group in self-reflectiveness on the BCIS (p= 0.047), total BCIS (p= 0.045), and intolerance to frustration (p= 0.014). Jumping to Conclusions (JTC) improved more in the MCT group than the psycho-educational group (p= 0.021). Regarding the comparison within each group, Theory of Mind (ToM), Personalizing Bias, and other subscales of irrational beliefs improved in the MCT group but not the psycho-educational group (p< 0.001–0.032).ConclusionsMCT could be an effective psychological intervention for people with recent onset of psychosis in order to improve cognitive insight, JTC, and tolerance to frustration. It seems that MCT could be useful to improve symptoms, ToM, and personalizing bias.