The study found that disruptive and rule-violating behaviors (criminogenic needs) and weak participation in treatment components (responsivity) were linked to an offender's failure to complete treatment. Variables that were not related to failure to complete treatment were sexual deviancy in itself, past employment history, education, intellectual abilities, cultural background, and marital status. Study participants were 64 adult male sex offenders (32 dropouts and 32 completers of treatment) who attended an intensive inpatient treatment program for high-risk sex offenders in an accredited forensic mental health facility in Canada between March 1995 and 2001. "Noncompleters" were defined as premature terminations from the program regardless of how close they were to program completion. This typically occurred only when problems escalated beyond reasonable and tolerable limits. Treatment attrition variables were grouped according to the risk, need, and responsivity dimensions. Discriminant function analysis was used to determine the accuracy of classifying completers and noncompleters. The authors argue that predictors of failure to complete treatment should not be used to exclude from treatment those that have those predictors, but rather they should be used to revise treatment components to address the causes of termination. 2 tables and 38 references