A stepwise multiple regression was conducted for 52 pedophiles serving sentences in a specialized prison for the treatment of sex offenders, with number of prior arrests for any/all felony offenses as the criterion variable; scores on the MMPI and the Limbic System Checklist, age of the victim in the instant offense, and age of the offender functioned as "post- dictor" variables. Eight variables were extracted, yielding a multiple R of .683. When the criterion is truncated into meta- categories (one or more arrests compared to none), a precision- weighted equation produced 90.4 percent "true positives" but only 9.6 percent "false negatives." The authors propose that the methodology presented in this paper, rather than the specific results obtained, can be extrapolated to other jurisdictions as a scientifically sound methodology that both reduces the rate of false negatives from that expected by chance and is likely to prove defensible in litigation. 5 tables and 38 references