Background: Suicide is a leading cause of death globally. Suicide deaths are elevated in those experiencing severe mental health problems, including schizophrenia. Psychological talking therapies are a potentially effective means of alleviating suicidal thoughts, plans, and attempts. However, talking therapies need to i) focus on suicidal experiences directly and explicitly, and ii) be based on testable psychological mechanisms. The Cognitive AppRoaches to coMbatting Suicidality (CARMS) project is a Randomised Controlled Trial (RCT) which aims to investigate both the efficacy and the underlying mechanisms of a psychological talking therapy for people who have been recently suicidal and have non-affective psychosis. Methods: The CARMS trial is a two-armed single-blind RCT comparing a psychological talking therapy (Cognitive Behavioural Suicide Prevention for psychosis [CBSPp]) plus Treatment As Usual (TAU) with TAU alone. There are primary and secondary suicidality outcome variables, plus mechanistic, clinical, and health economic outcomes measured over time. The primary outcome is a measure of suicidal ideation at 6 months after baseline. The target sample size is 250, with approximately 125 randomised to each arm of the trial, and an assumption of up to 25% attrition. Hence, the overall recruitment target is up to 333. An intention to treat analysis will be used with primary stratification based on National Health Service (NHS) recruitment site and antidepressant prescription medication. Recruitment will be from NHS mental health services in the North West of England, UK. Participants must be 18 or over; be under the care of mental health services; have mental health problems which meet ICD-10 non-affective psychosis criteria; and have experienced selfreported suicidal thoughts, plans, and/or attempts in the 3 months prior to recruitment. Nested qualitative work will investigate the pathways to suicidality, experiences of the therapy, and identify potential implementation challenges