Offenders with mental illness have attracted substantial attention over the recent years, given their prevalence and poor outcomes. A number of interventions have been developed for this population (e.g., mental health courts). They share an emphasis on one dimension as the source of the problem: mental illness. Their focus on psychiatric services may poorly match the policy goal of reducing recidivism. In this article, we use research to evaluate (a) the effectiveness of current interventions, and(b) the larger viability of psychiatric, criminological, and social psychological models of the link between mental illness and criminal justice involvement. We integrate theory and research to offer a multidimensional conceptual framework that may guide further research and the development of efficient interventions that meaningfully reduce recidivism. We hypothesize that the effect of mental illness on criminal behavior reflects moderated mediation (i.e., the effect is direct in the case of one subgroup, but fully mediated in another); and that the effect of mental illness on other "recidivism" is partially mediated by system bias and stigma. We use this framework to propose three priorities for advancing research, articulating policy, and improving practice.