In their economic modelling, Barret & Byford ^{Reference Barrett and Byford1} postulate that the intervention group will have a reoffending rate of 3% v. 5% in the non-intervention group, but give no evidence of this being the correct figure or even the justification for this being a reasonable estimate. It is possible that the authors are assuming that the protective effects of being in detention and receiving treatment as part of the dangerous and severe personality disorder (DSPD) programme reduces the risk to the public more than being released into the community. However, this protective effect may just be down to being detained, whether receiving treatment or not. In any case, for cost-effectiveness there has to be a justification for the effectiveness figures used, and none was presented in the paper. It is clear that in the modelling the best option is to be detained in a low-cost prison and the authors should have modelled the possibility of the therapeutic part of the DSPD programme having limited effect over detention, i.e. that it is the preventative detention effect that is important not the therapeutic part. The authors provide further evidence that the best management of violent offenders is for the criminal justice system to manage risk by protecting the public by keeping dangerous offenders in prison for long periods. There does not seem to be an economic reason to place these patients on a mental health treatment programme with, so far, unknown efficacy but high costs. The health pound would better be spent in evidence-based treatment programmes for mental illness instead.