Summary Background Mortality in patients in psychiatric hospitals is reported to be two to three times as high as in the general population. In Tunisia, we do not have any figures on mortality and causes of death in psychiatric inpatients. Aim The aim of our study was to assess the mortality rate in a psychiatric hospital in comparison to the mortality rate in the general population, to determine the patients’ profile, and to identify the causes and risk factors for these deaths. Methods We performed a retrospective, descriptive and comparative study. We examined the records of all patients who died during their stay in the different wards of psychiatry at the Razi Hospital in Tunis. We also scrutinized reports of autopsies in the Forensic Medicine unit at Charles-Nicolle Hospital in Tunis over a period of eleven years from January 1st, 2000 to December 31st, 2010. We conducted a descriptive study to calculate the standardized mortality ratio (SMR) aiming to highlight any existing excess mortality among the psychiatric inpatients compared to the general population. This ratio was obtained by dividing the observed number of deaths by the expected number of deaths. In the analytical study, our sample was compared to a control population made-up of randomly selected living patients among those admitted to the Razi hospital in 2010. This study allowed us to investigate the risk factors for premature mortality in psychiatric inpatients. Results The average rate of mortality was two deaths per 1000 inpatients per year. Twenty-four percent (24 %) of deaths involved institutionalized patients. Compared to the general population, premature mortality was noted among patients aged less than 40 (SMR=1.9). The older the patients were, the closer to 1 the SMR was. The average age at death was 51.38 years; 65 % of patients were male, 60 % had a low socio-economic level, 54 % had a comorbid medical condition. Forty-two percent (42 %) of deceased patients were diagnosed with schizophrenia with the paranoid form being the most prevalent (44 %), 13 % had bipolar disorder, 22 % had psycho-organic disorders (mental retardation, dementia, delirium). Antipsychotics were the most prescribed psychotropic drugs. High doses were used. Forty percent of cases (40 %) consisted of sudden deaths. A cause for death was identified in 80 % of cases. In 92 % of cases, the death was classified as being “natural”. Main causes were respiratory (26 %) and cardiovascular (9 %). Accidental causes accounted for 8 % of deaths. In 20 % of cases, the cause remained undetermined. Three factors were identified as independent predictors of mortality among mental patients: age at death (OR=3.9 among patients older than 40), psychiatric diagnosis (OR=2.9 among patients with psychotic or mood disorders compared to other diagnoses) and combination of antipsychotic drugs (OR=6.09 in patients receiving more than two antipsychotics). Discussion Young psychiatric inpatients seem to be at high risk of premature death: the SMR in our study was 1.9. It ranged between 2.15 and 6.55 in other similar studies. This increased risk mainly concerns non-natural deaths. The leading natural cause of death in our population was represented by thromboembolic accidents. Such a high thromboembolic risk may be explained by the mental illness itself, by physical restraint as well as by antipsychotic treatment. Diagnosing medical conditions in psychiatric patients is often a daunting task: history of the patient is sometimes unreliable and clinical features might be modified by psychotropic agents. Patient-related risk factors for premature death include poor socio-economic level, access-to-care difficulties, positive family and personal history of mental and/or medical disorders, smoking, substance abuse, unhealthy diet and lack of physical activity. Moreover, iatrogenic effects of psychotropic drugs (combination of antipsychotics was more common in deceased patients than in controls) and inadequate medical care in psychiatric hospitals (lack of ECG devices, in particular) partly account for such a high mortality. Conclusion Identifying risk factors for deaths in psychiatric hospitals highlights needed changes in psychiatric management strategies taking into account the patient's characteristics as well as the drugs’ safety profile. Further studies with larger samples are needed to better highlight risk factors for premature death in psychiatric inpatients. Identifying such risk factors is necessary to develop efficient preventive strategies.