BackgroundThe current study is a secondary analysis of qualitative data collected as part of EURIPIDES, a study which assessed how patient experience data were used to improve the quality of care in National Health Service (NHS) mental health services.ObjectiveWe undertook a detailed realist secondary qualitative analysis of 10 interviews in which expressions of racialisation were unexpectedly reported. This theme and these data did not form part of the primary realist evaluation.MethodsInterviews were originally conducted with the patients (18–65 years: 40% female, 60% male) from four different geographically located NHS England mental health trusts between July and October 2017. Secondary qualitative data analysis was conducted in two phases: (1) reflexive thematic analysis and retroduction; (2) refinement of context–mechanism–outcome configurations to explore the generative mechanisms underpinning processes of racialisation and revision of the initial programme theory.FindingsThere were two main themes: (1) absence of safe spaces to discuss racialisation which silenced and isolated patients; (2) strained communication and power imbalances shaped a process of mutual racialisation by patients and staff. Non-reporting of racialisation and discrimination elicited emotions such as feeling othered, misunderstood, disempowered and fearful.ConclusionsThe culture of silence, non-reporting and power imbalances in inpatient wards perpetuated relational racialisation and prevented authentic feedback and staff–patient rapport.Clinical implicationsRacialisation in mental health trusts reflects lack of psychological safety which weakens staff–patient rapport and has implications for authentic patient engagement in feedback and quality improvement processes. Larger-scale studies are needed to investigate racialisation in the staff–patient relationships.