The proliferation of indecent images of children (IIOCs) on the Internet has exceeded the resources required to investigate suspects effectively. This article examines the validity of the Kent Internet Risk Assessment Tool—Version 2 (KIRAT– 2), an evidence-based framework for prioritizing IIOC suspects according to their risk of committing contact offenses against children. Data were obtained from 374 police files (11 police forces across the United Kingdom) corresponding to individuals who were convicted for IIOC offense(s) during the period 2001–2013. Some 170 offenders had convictions or allegations of contact sexual offenses against children; police defined these as higher risk of contact offending (HR) as opposed to the remaining 204 individuals without convictions or allegations of contact sexual offenses (lower risk of contact offending, LR). We coded 166 variables previously discussed in the literature or from law enforcement experience that may discriminate dual from noncontact offenders. Receiver operating characteristic (ROC) curve analysis was conducted to test which variables or combination of variables discriminated between HR and LR offenders. The final model, represented as a phased decision tree, uses 17 variables with 4 filters or decision steps examining previous convictions, access to children, current evidence of both online and offline behavior, and other relevant factors. The model classified 97.6% of HR within the higher risk levels (high or very high) and 62.3% of LR within the lower risk levels (low or medium). Findings are discussed in terms of contribution to the Internet sex offending risk assessment literature and practical implications for police forces.