Introduction: Evaluation of variables correlated to homicide is a fundamental issue for developing preventive and therapeutic strategies to deal with such criminal behavior. Objectives: The objectives of this study were to assess the characteristics of homicide in Tunisian patients suffering from schizophrenia and to determine the correlated socio-demographic, clinical and therapeutic variables. Methods: The study included two groups of male patients with a DSM-IV diagnosis of schizophrenia who attended the 'Razi' university psychiatric hospital of Tunis. The first group was composed of 36 patients hospitalized for homicide in the forensic unit between the first of January 2000 and the 30th of May 2012. The second group included 50 patients without any criminal record. Demographic, clinical and therapeutic variables were analyzed and compared between the two groups. Results: No differences were found between the two groups regarding the different socio-demographic variables. Significant differences were found with respect to a duration of untreated psychosis equal to or more than one year (p = 0.048), shorter duration of psychiatric care (p = 0.002), lower number of hospitalizations (p = 0.026), antecedent of forced hospitalization (p < 0.001), low degree of insight (p = 0.001), poor medication compliance (p < 0.001) and higher antipsychotic doses (p = 0.001). Discussion: Demographic variables as suggested by other studies are less valuable predictors of homicide in patients with schizophreniConclusion: Interventions for reducing such behavior should focus on clinical variables and integrate an early diagnosis of the disease and improvement of insight as well as medication compliance.