Overall, the findings suggest that the SORAG may be a more accurate predictor of sexual recidivism compared to the Static-99 among samples of higher risk sex offenders. The risk percentages may not be valid for higher risk offenders who have completed an effective sex offender treatment program because of the greater treatment effects. Of the 258 offenders in the sample, 23 (8.9 percent) reoffended sexually over an average of 5.1 years of followup. For violent (including sexual) reoffending, 34.1 percent of the sample reoffended over a 4.6-year followup period. The SORAG was found to have moderate predictive accuracy for both sexual and violent reoffending over a 5-year followup period, but the Static-99 was accurate in predicting only sexual reoffending. When the published risk percentages for each instrument were compared with the actual reoffending rates in the current sample, the current sample reoffended at a lower rate than the expected rate. This may be due to the finding of Andrews and Bonta (2003) that the most significant treatment effects are likely to occur for high-risk rather than low-risk offenders. The majority of the current sample was in the higher risk categories. Subjects were drawn from a sample of 380 consecutive admissions to the Regional Treatment Center Sexual Offender Treatment program, a residential, 7-month treatment program housed in a maximum-security psychiatric hospital that serves the Ontario Region of the Canadian Federal prison system. All had scores on the Static-99 and the SORAG. Of the 380 sex offenders, 258 had been released at the time of followup (September 2004). Analyses were limited to this latter subsample of offenders. 5 tables and 26 references