Black and minority ethnic (BME) people using psychiatric services are at greater risk of non-engagement, dropout from care and not receiving evidence-based interventions than white British people. To identify effective interventions designed to improve therapeutic communications (TCs) for BME patients using psychiatric services in the UK, to identify gaps in the research literature and to recommend future research. Black African, black Caribbean, black British, white British, Pakistani and Bangladeshi patients in psychiatric services in the UK, or recruited from the community to enter psychiatric care. Some studies from the USA included Hispanic, Latino, Chinese, Vietnamese, Cambodian and African American people. Any that improve TCs between BME patients and staff in psychiatric services. The published literature, 'grey' literature, an expert survey, and patients' and carers' perspectives on the evidence base. Databases were searched from their inception to 4 February 2013. Databases included MEDLINE, Applied Social Sciences Index and Abstracts, The Cochrane Library, Social Science Citation Index, Allied and Complementary Medicine Database, PsycINFO, Cumulative Index to Nursing and Allied Health Literature, EMBASE, The Campbell Collaboration and ProQuest for dissertations. Studies were included if they reported evaluation data about interventions designed to improve therapeutic outcomes by improving communication between BME patients and psychiatric professionals. Qualitative studies and reports in the grey literature were included only if they gave a critical evaluative statement. Two members of the team selected studies against pre-established criteria and any differences were resolved by consensus or by a third reviewer, if necessary. Data were extracted independently by two people and summarised in tables by specific study designs. Studies were subjected to a narrative synthesis that included a thematic analysis contrasting populations, countries and the strength of evidence for any intervention. The components of the interventions were compared. Patient perspectives on acceptability were considered alongside quality scores and methodological strengths and weaknesses. Twenty-one studies (19 from the published literature and two from the grey literature) met the inclusion criteria. There were 12 trials, two observational quantitative studies, three case series, a qualitative study and three descriptive case studies. Only two studies, one a pilot trial and one a case series, included economic data; in both, a favourable but weak economic case could be made for the intervention. The trials tested interventions to prepare patients for therapeutic interventions, variable levels of ethnic matching (of professional to patient), cultural adaptation of therapies, and interventions that included social community systems in order to facilitate access to services. Empowering interventions favoured by patients and carers included adapted cognitive-behavioural therapy, assessments of explanatory models, cultural consultation, ethnographic and motivational interviews, and a telepsychiatry intervention. Studies tended to have small sample sizes or to be pilot studies, and to use proxy rather than direct measures for TCs. Empowering interventions should be further researched and brought to the attention of commissioners. Several promising interventions need further evaluative research and economic evaluations are needed. The study is registered as PROSPERO CRD42011001661. The National Institute for Health Research Health Technology Assessment programme.