AbstractA probation risk‐assessment instrument developed by the State of Wisconsin is receiving wide use throughout the nation and has been recognized by the National Institute of Corrections as part of a “model system.” However, this has been done without extensive validation on populations other than Wisconsin probationers and parolees. The validity of the instrument for a population of City of New York probationers is assessed in this study. It was found that many of the variables contained in the instrument did not predict risk for the sample. On this basis. the validation of risk instruments before they are fully adopted is recommended. In addition, based on an analysis of split halves of our sample, the general weakness of statistically derived models is