In the United States, randomization is viewed broadly to be the best method of gaining knowledge on the causal effects of treatments in the social sector, including legal measures in general and sanctions of criminal law. Randomized experiments are rare in the social sciences in Germany, on the other hand, and no guidelines on their use exist. In German criminal law, experimentation has been regarded as absolutely forbidden. More recently, however, there has been strong pressure to conduct controlled experiments that benefit the offender, which means testing the effectiveness of sanctions less severe than the status quo and comparing them to it. This addresses the issue of treating offenders with sanctions of doubtful effectiveness. Lately, the example of experimentation on social therapy in prison (Ortmann, 1994) proved that randomization in German criminal law is possible in practice, even though the theory might not have changed. This article discusses whether such a change should be made and whether the debate in Germany can be improved by learning from the United States' experience with experimentation. The guidelines for balancing the risks of experimentation against their supposed benefits as developed by the United States' Federal Judicial Center are measured against the German constitution. The author argues that unequal treatment in randomized designs can only be justified if randomization is permitted for reasons other than experimentation and if the individuals give their free and informed consent. 18 references