This pilot study examined differences in criminal attitudes and hostile attribution biases among three groups of offenders diagnosed with serious mental disorders. It also compared the criminal outcomes for a sample of mentally ill offenders (MIO's) who completed a modified version of the Options cognitive skills program with a treatment-as-usual sample of MIO's. Results showed that pre- and posttest change scores for the Overgeneralization subscale of the Hostile Interpretations Questionnaire (HIQ), the Identification with Criminal Others subscale of the Criminal Sentiments Scale—Modified (CSS-M), and the HIQ Authority subscale differed between the groups of offenders with serious mental disorders. Participants identified for treatment by the pilot program (intended treatment group) and participants who completed the entire intervention had fewer arrests, including fewer arrests for violent offenses, but had higher technical probation violations than a treatment-as-usual comparison group.