The Level of Service Inventory-Revised (LSI-R) was developed by Andrews and Bonta in 2001. It consists of static and dynamic variables and provides information about an offender's risk of recidivism and about the need for treatment and supervision.

In the past, most validation studies analyzed rather short follow-up periods and focused on non-violent offenders serving short sentences. Studies that investigated the predictive validity for offenders serving long-term sentences found a moderate performance.

The purpose of the present study was to show whether the LSI-R is predictive for recidivism of offenders in Switzerland and whether it qualifies for short-term prognosis (one year) as well as long-term prognosis (seven years).

The sample consisted of 107 violent and sex offenders. Follow-up time was seven years.

The ROC-analyses showed a moderate predictive validity for recidivism within one year (AUC=0.66) and a good predictive validity for recidivism within seven years (AUC=0.78).

The results suggest that the inclusion of dynamic items does not necessarily limit the predictive validity of an instrument to short follow-up periods or to specific criminal justice systems or cultural settings.