Résumé : The aim of the study was to explore the relationships between subjective or objective symptoms and mortality in schizophrenia. 310 subjects meeting the ICD-10 criteria for schizophrenia were included in the study between 1998 and 2000. At the initial assessment the following variables were respectively assessed to evaluate subjective and objective symptoms: the Frankfurt Complaints Questionnaire (FCQ) and the Positive and Negative Syndrome Scale (PANSS). In May 2008, information about the subjects were collected in order to know if they are alive or not and if they are deceased to know the date and the causes of their death. Survival analysis was conducted using the Kaplan-Meier product-limit estimator and standardized mortality ratio (SMR) was calculated. A multivariate Cox regression was done to detect predictive factors associated with mortality. Absolute mortality rates were 10.01%, 4.46% and 5.42% for overall mortality, unnatural causes and natural causes, respectively. SMR for overall mortality was 4.73. Cox regression analyses showed that elevated scores of FCQ was significant predictor of deaths from unnatural causes. High levels of subjective symptoms, as rated by the FCQ were independent predictor of mortality by unnatural causes in schizophrenic subjects. There were several limitations: The causes of death were not determined by autopsy and secondly, the duration of the study could be insufficient to detect significant associations between clinical variables and mortality.