These notes place an important limitation on the conclusions drawn by Friesen and Andrews (1982) regarding the predictive validity of a measure of the inprogram self-regulation efforts of 42 young-adult probationers. In the earlier report, assessments of self-regulation were found to be reliable (interrater r(20) = .96) and they correlated with intake socialization scores ( r(42) = .38), and with reconvictions monitored from the start of probation to the end of three postprobation years ( r(42) = -.38). The present note shows that the predictive validity of the self-regulation scores was only evident among high-risk cases: The correlation was -.70 with the recidivism of 19 low-socialization cases compared with -.02 among 23 high-socialization probationers. The results are discussed in relation to the risk principle of case classification.