Background: Numerous studies have shown that the quality of the therapeutic relationship (TR) between the patient and the clinician is an important predictor of the outcome of different forms of psychotherapy. It is less clear whether the TR also predicts outcomes of psychiatric treatment programmes in patients with psychosis (i.e. outside conventional psychotherapy). Methods: We conducted a systematic review and identified 9 primary studies that prospectively tested the association of the TR with 3 outcomes, i.e. hospitalisation, symptom levels and functioning. Because of the heterogeneity of the methods used, a meta-analysis was not feasible. A vote counting method was used to determine the number of statistically significant effects in the hypothesised direction (i.e. that a more positive TR predicts more favourable outcomes). Results: For each outcome, a χ2 analysis showed that the number of statistically significant findings in the hypothesised direction was greater than expected if the null hypothesis of no association were true. However, studies had methodological shortcomings, and the effect sizes of positive associations were rather small. Conclusion: It may be concluded that there is some, but not overwhelming, evidence that the TR predicts outcomes of complex psychiatric treatment programmes in patients with psychosis, and that methodologically more rigorous research is required. Such research should measure the TR at initial stages of treatment and use validated assessment instruments for both TR and outcomes.