The National Research Council (NRC) report on Improving Evaluation of Anticrime Programs presents and discusses a wide array of techniques of evaluation. Although recognising the very high internal validity of randomised experiments, it considers, under certain conditions, quasi-experiments and observational studies as equally valid approaches. This conclusion is critically reviewed from a European perspective, where only a few randomised trials have been realised so far. It is argued that many critiques routinely addressed to randomised experiments, such as ethical concerns or low acceptance among practitioners, are either unfounded or can be adequately dealt with through imaginative adjustments. On the other hand, randomised controlled trials need to take the challenge of broadening the perspective, especially by looking at long-term effects that no other method can consider with comparable internal validity. Other recommendations include using innovative measures of re-offending, considering dynamic rather than static criteria of re-offending, and looking, beyond re-offending, at rehabilitation in other areas of life. Particular challenges are the possible placebo effects that evaluators in criminal justice have not yet found appropriate ways to deal with.