This paper addresses a simple question: what have researchers in the United States and other countries learned about probation performance generally and the effectiveness of specific probation practices in particular? While the 䳣ience? derived from the evaluation studies is still weak, it has been argued that probation could be organized along three risk dimensions, targeting high-risk times, high-risk offenders and high-risk locations. Research examining these risk dimensions is presented here, and the implications for an emerging desistance-based probation paradigm are considered