A clear structured approach to violence risk assessment that is both, evidence-based and gender specific- is high on the political and mental health agendas. The individual risk of perpetrators depends on several parameters that are incorporated into assessment instruments. Most data about risk factors included in risk assessment instruments were derived only on male offenders.

This study is part of Germany's biggest risk assessment study, -the Munich Prognosis project (MPP) - and focuses on factors included in risk assessment instruments associated with criminal and violent recidivism in a sample of male and female delinquents referred for forensic-psychiatric evaluation prior sentencing. The predictive validity of four instruments (HCR 20, ILRV, VRAG and PCL-R) for violent and general repeat offenses was analyzed.

When comparing the predictive validity of the four instruments for male offenders, the results were in favor of the PCL-R, i.e. PCL-R Factor 2 when focusing on violent recidivism.

For female offenders ROC analysis found superior results of the HCR 20-R items (AUC .793 p< .05), the ILRV D variables (AUC .814), p< .05) and the VRAG (AUC .864, p< .05) for violent recidivism. They were in favor of the PCL-R Factor 1 (AUC .666, p< .05) when focusing on general recidivism.

The importance of gender specific violence risk assessment will be discussed.