This article is concerned with the question how the risk of reoffending in mentally disordered offenders can be predicted. First, an outline is given of the Dutch legal and therapeutic context calling for such a prediction, the judical treatment-measure called TBS. Then an attempt is made to apply previous research in the field of predict- ing the risk of reoffending to TBS. Due to vagueness both about the target group for prediction as well as the definition of the behavior to be predicted, most of this re- search appears to be hardly valid within the Dutch situation. Moreover, those variables actually found to have predictive value are mostly of a static nature and therefore not susceptible to treatment.

In the second half of the article results are presented from research by the authors, based on a questionnaire which aims at the prediction of the risk of reoffending. It appeared possible to isolate five variables that are both susceptible to treatment and significantly linked to actual recidivism.