Background The reasons behind suicide are multifactorial, complex, and poorly understood. Despite decades of research, suicide rates remain elevated both during hospitalization as well as shortly after discharge. Clinically relevant knowledge and tools for suicide risk assessments to guide decision-making processes in acute settings are needed. Objectives The thesis objective was to examine and improve understanding of the complex relationship between psychotic depression and suicide behaviour. We also examined risk factors of clinical relevance in hospital settings, both during treatment, and after discharge. Methods This research project used a sequential exploratory strategy, commencing with a qualitative part (paper 1-2), followed by a quantitative part (paper 3-4). In paper 1 and 2, we conducted qualitative interviews with nine patients diagnosed with psychotic depression as part of uni- or bipolar disorder. Papers 3 and 4 used data from the naturalistic prospective cohort study Suicidality in Psychiatric Emergency Admissions. In paper 3, we included all patients with depression (N= 1846) in a representative sample to identify risk factors for suicide using a cox regression analysis. Paper 4 investigated a mixed diagnostic sample of 7000 acutely admitted inpatients to identify imminent- and short-term risk factors for suicide. The findings from all four papers were combined into a secondary analysis, where we thematically synthesized the perspectives of patients’ experiences with the main findings from the prospective cohort study based on clinical assessment. Results thematic summarized Our main finding was that psychotic symptoms in depression are associated with increased suicide risk, both short- and long-term (Paper 3). A possible mechanism to explain this association is the combination of intense emotional suffering and cognitive impairment which may motivate impulsive behaviour in psychotic depression (Papers 1 and 2). Depressed mood including overvalued ideas and delusions of self-blame and guilt predicted suicide the first following week after admission (Paper 4). The association of intense affective states with both imminent and long-term suicide risk factors was supported by patient experiences of sleeplessness and intense anxiety as warning signs for suicidal behaviour (Paper 1 and 2). In further support of this, depression severity assessed by the Health of the Nation Scale item 7 predicted suicide the first following week after admission (Paper 4). We found higher short-term suicide rates in patients diagnosed with psychotic depression as compared to those without psychotic symptoms at all follow-up points (Paper 3). A diagnosis of psychotic depression and depression severity assessed by MADRS depression scale as independent predictors of suicide provided knowledge on the long-term perspective (median 5.5 years follow-up). Also of clinical importance is our finding that self-report of suicidal ideation is not a good measure of imminent, short- or long-term suicide risk. Our qualitative findings indicate that shame, psychotically motivated paranoid ideation and impulsivity motivated underreporting of both suicidal ideation and psychotic symptoms in themselves. This may partly explain why, despite reporting less suicidal ideation/planning on admission, the suicide rate was higher for patients with psychotic symptoms on admission than for those without such symptoms (Paper 3). Although reported suicidal ideation was less relevant for clinical risk assessment, paper 3 and 4 identified male gender and recent suicide attempt as significant risk factors for completed suicide. The last main finding was that patients found both security measures and a treatment approach focusing on modifying psychotic experiences, intense anxiety and sleeplessness was helpful during hospitalization. Taking the time to talk to establish good patient relations is essential for optimal treatment conditions. Conclusions The periods of acute inpatient treatment and immediately post- discharge are associated with the highest suicide risk. High symptom load as well as a diagnosis of psychotic depression, especially the presence of overvalued ideas and delusions of self-blame and guilt, appear to indicate especially high suicide risk regardless of time horizon. Identifying delusions and other psychotic symptoms is thus of the utmost importance. Psychosis- and depression-focused interventions and treatment approaches aiming specifically to reduce both psychotic and affective symptom load may act as a suicide prevention strategy for psychotic depressed patients. We found self-report of suicidal ideation of limited relevance when assessing suicide risk. Such assessments should thus always be combined with thorough clinical assessment by trained staff. This is especially true when assessing patients with psychotic depression, but especially given the frequent underreporting of psychotic symptoms, it remains important with any patient admitted for a depressive episode. Structured depression rating scales for clinician use may meaningfully contribute to suicide risk assessments. When assessing acute admission patients, we recommend paying particularly close attention to males, and patients with prior and recent suicide attempts.