Research on risk assessment with offenders with an intellectual disability (ID) has been scarce; the overwhelming majority of risk tools available are developed for mainstream populations. This thesis is primarily concerned with understanding static and dynamic risk factors for physical and sexual violence among offenders with an ID. This thesis described a series of quantitative and qualitative studies investigating the utility and predictive accuracy of risk assessments among this population and culminated in the development of a new ID focused risk tool. In Chapter 1, a brief introduction outlined current research and practice regarding risk assessment and prediction in the ID field and identified significant limitations in the evidence base. In Study 1 (Chapter 2) a dynamic risk assessment (ARMIDILO-S) for sexual offenders with an ID resulted in the best prediction of sexual reoffending when compared with established static risk assessment tools (STATIC-99 and VRAG) developed for mainstream offenders. Study 2 (Chapter 3) adopted a public health model of understanding how static and dynamic risk factors ‘work together’ to predict violent behaviour. The findings suggested that the two approaches essentially measure similar underlying risk which has important implications for the future of risk assessment procedures with this population. Offenders with an ID were the focus of a qualitative study (Chapter 4) in which it was found that environmental factors featured heavily in the participant’s explanations of their own aggressive behaviour. The final empirical study (Chapter 5) details the construction and initial validation of a new dynamic risk measure: Current risk of Violence (CuRV). The CuRV demonstrated promising reliability and validity as an assessment of aggression. Finally, in Chapter 6, findings from the four empirical studies were discussed in relation to their contribution to the literature, theoretical and clinical implications, methodological limitations, and potential avenues for future research.