Violent offender rehabilitation programs aim to reduce the risk of re-offending in known offenders by addressing a range of different treatments needs, often with core intervention targets of improving anger regulation and altering antisocial beliefs and thinking styles. Such programs have proven efficacy in reducing recidivism for some, but not all, violent offenders, and little is known about the effects of these programs on different offender types. This study investigates whether subtypes of violent offenders can be meaningfully identified and considers how this influences short-term treatment outcomes. Cluster analysis identified three distinctive violent offender groups within a sample of 305 male offenders who had been assessed for participation in a violent offender rehabilitation program. An "unregulated" group had high levels of anger experience and expression and low levels of anger control, and held beliefs that were strongly supportive of a criminal lifestyle. A "regulated" group demonstrated levels of anger and beliefs supporting criminal activity that were not in a range that warranted treatment. Finally, an "overregulated" group was assessed as the group at highest risk of violent re-offending and had low levels of anger experience and expression and an absence of beliefs supporting criminal activity. The unregulated group appeared to gain the most benefit from treatment, although it had the highest levels of criminal thinking and problematic anger. These findings nonetheless offer support for the hypothesis that violent offender treatment programs may be optimally effective when targeted at particular types of offenders.