Background: Research on prediction of violent and sexual offending behaviour has developed considerably in the mainstream criminological literature. Apart from one publication [Quinsey (2004) "Offenders with Developmental Disabilities," pp. 131-142] this has not been extended to the field of intellectual disabilities. Methods: Work on actuarial instruments, dynamic variables, clinical judgement and structured clinical judgement is reviewed. Results: A number of studies comparing actuarial instruments in terms of their predictive validity are reviewed. Relative effectiveness and applicability to intellectual disability is considered. A framework for dynamic variables is outlined and the importance of dynamic variables for inclusion in risk prediction is established. Strengths and limitations of clinical judgement are reported and the importance of reliability is noted. Finally, structured clinical judgement is reviewed in terms of the way in which it combines the other three groups of variables. Conclusions: The information regarding different methods of risk assessment is integrated with research and opinion contained in the Special Issue. Risk prediction will always be a judgement and as such there will always be errors in judgement. As clinicians, researchers and policy makers it is our duty to employ the latest research information to make predictions that are as accurate as possible. However, we must also help to promote a culture that can be tolerant of inevitable errors.