Risk of violence among offenders who have intellectual disability (ID) is a problem in forensic psychology and criminal justice. Conventional risk assessment instruments are usually inefficient in responding to the dynamic aspects of risk factors relating to this group of people, whereby the circumstances may vary and therefore influence the violent behavior. The paper will describe a dynamic risk assessment model that aims to increase the level of accuracy in predicting violence amongst intellectually disabled offenders as it attempts to combine both the predictors that are considered as dynamic (i.e., mood, stress, social interactions) and the ones that are considered as static (i.e., age, criminal history). Information on 300 intellectually disabled offenders was gathered, consisting of historical data, psychological testing, and reports on behaviors. The model has been created with the help of statistical techniques and has been confirmed with the cross-sectional analysis and longitudinal analysis with the help of regression analysis and machine learning tools like decision trees and random forests. The estimation of the relationship between the risk factors and violent behavior was done by logistic regression. The dynamic model was found to have much better predictive validity than its traditional counterpart, the static model. The important results are that predictive accuracy (72 % to 82 %) improved by 30 %, and it was possible to identify high-risk offenders better. The model has high sensitivity rates of 85 % and specificity of 78 %, and it has been identified as a very efficient model in determining both those who are violent and those who are not. AUC of the dynamic model was 0.88 better than that of the static models (AUC: 0.75). Other factors that were found to be critical dynamics in the model that are involved in the escalation of violence include social isolation and environmental stressors. The results indicate that dynamic models have the potential to enhance violence prediction and management in intellectually disabled offenders. Further validation, investigation of real-time data combination, and its use in alternative criminal justice systems should be addressed in future research.