Prisons are often seen as places where violence is generally accepted and frequent. Incidents in prisons jeopardize effective functioning of prisons, while victims of violence have problems both with adjustment to conditions of imprisonment and with inclusion to community after release. Participation in incidents during imprisonment is perceived as one of the most significant predictors of future criminal behaviour (recidivism). Causes of prison violence are explained by different theories which are then tested in correctional settings. Most common theoretical models to explain violent behaviour in prison are importation, deprivation and situational model. Importation model explains prison violence through personal characteristics, values and norms of offenders which are then brought into prison. In such terms, predictors of criminal behaviour are also predictors of negative forms of behaviour in prisons. Deprivation model finds explanation of prison violence in characteristics of prison as (total) institution and deprivations arising from imprisonment, where prisoners’ subculture represents a form of confronting formal control and reducing deprivations. Situation model explains human behaviour through interaction between personal and environmental characteristics. According to this explanation, prison as institution is focused on gaining control, where strategies to achieve control comprise elements of environment and elements of interaction between prison officials and prisoners. As prison violence is a phenomenon which provokes scientific and expert interest, numerous studies are trying to determine causes and consequences of violence and possibilities of effecting violence and incident behaviour in prisons. Two aims were set to explore phenomenon of violence in Croatian prison system. First aim was to establish personality traits, behavioural and criminological characteristics and differences in these characteristics between male and female prisoners with disciplinary measure declared for committing disciplinary offence. Second aim was to establish predictive value of importation, deprivation and situational model in explaining violent behaviour of male and female prisoners in Croatian prison system. Two samples were formed to conduct research. First sample (N=173) was constituted of male prisoners from Penitentiary in Lepoglava (NM=126) and female prisoners from Penitentiary in Požega (NF=47) with disciplinary measure declared for committing disciplinary infraction. Second sample was constituted of disciplinary infractions (N=364) for which disciplinary measures were declared. To process data for the purpose of the first and the second aim of the research methods of descriptive analysis (absolute and relative frequencies, measures of central tendency), t-test and hi-square test were used. In the scope of the first aim discriminative analysis was applied to determine differences in characteristics between male and female prisoners who committed disciplinary offences. For the purpose of second aim - determining predictive value of importation, deprivation and situational model of violent behaviour separately for male and female prisoners, logistic regression analysis was applied. Research found differences in personality traits, behavioural and criminological characteristics between male and female prisoners, where adverse characteristics in all three categories were found for men. Within personality traits, main characteristic to differ male from female prisoners is aggressiveness - male prisoners are more aggressive then female prisoners. In the category of behavioural characteristic male and female prisoners differ mostly in alcohol consumption - male prisoners express significantly more problematic attitudes towards alcohol consumption then female prisoners and they are more likely to have problem with alcoholism and excessive drinking. Criminological characteristics to differ male from female prisoners the most are length of the sentence, admitting criminal offence and previous convictions - male prisoners have longer sentences, are less likely to admit criminal offence and are more likely to be recidivists. Research confirmed predictive value of importation model in explaining occurrence of violent disciplinary infractions committed by male and female prisoners. Also, predictive value of some individual variables of the importation model was confirmed, but only for prisoners - number of children and aggressiveness. Prisoners with more children are more likely to commit violent disciplinary infractions and average or above-average level of prisoners’ aggressiveness has predictive value for committing violent disciplinary infractions. Predictive value of deprivation model was found only for female prisoners and not for male prisoners, but none of the individual variables of deprivation model showed predictive value in explaining violent behaviour of female prisoners. Predictive value of situational model was found only for male prisoners, but not for female prisoners. Individual variables of situational model which showed predictive value refer to time when disciplinary infraction was committed - as a rule disciplinary infractions do not occur in free time and during weekends but in working days and working time. Results indicate specificity of Croatian prison system in relation to determined predictors of violence, necessity of gender specific approach to violence in the prison system and recommendation to implement future research of predictors and correlates of violence on larger sample, targeting specific types of violence and by using combined research approach (quantitative/qualitative).