Research examining the efficacy of cognitive behavioral therapy (CBT) in reducing recidivism has paid little attention to treatment factors contributing to response variability. Using an archival sample of 448 participants exposed to a risk-need-responsivity (RNR)-informed CBT program or no treatment, a multigroup latent profile analysis yielded a four-profile solution: a treatment-nonresponsive group and three treatment-responsive groups. Among the treatment-responsive profiles, reduced criminal attitudes were most predictive of desistance from reoffending. Elevated rates of recidivism and negligible gains following treatment were associated with pretreatment elevations in antisocial traits, risk level, and negative attitudes toward treatment. These findings underscore a greater need for individualized assessment of risk and treatment motivation, the importance of altering criminal sentiments to prevent reentry into the system upon release, and challenge the idea that 200 hours of treatment is sufficient for lasting change. Study limitations and further directions are discussed, including the need for correctional treatment outcome research to better isolate individual differences.