Purpose: Research on dynamic risk assessment has developed over the last 10 years and a number of variables have emerged as being possible predictors of future sexual and violent offences. These variables include hostile attitude/anger and compliance with routine. In 2002, Thornton ("Sexual Abuse: A Journal of Research & Treatment" 14, 139) developed a framework for dynamic risk assessment which also includes distorted and dysfunctional attitudes, socio-affective functioning, self-management and, for sex offenders, sexual interests. In 2000, Hanson & Harris ("Criminal Justice & Behaviour" 27, 6) provided empirical support for the predictive power of a number of dynamic factors which they split into stable and acute variables. In their research they found poor social supports, antisocial lifestyle, and poor self-management strategies to be important in addition to co-operation with supervision and hostile attitude. The present report describes the development of the Dynamic Risk Assessment and Management System (DRAMS)--an assessment for dynamic/proximal risk factors in people with intellectual disabilities. Method: In a field trial, the DRAMS was administered by staff unfamiliar with the formal requirements of the assessment. They were familiar with other ward-based assessments. Ten subjects were assessed for reliability purposes (total of 45 assessments) and five subjects were assessed consistently over a 3-month period. Eighteen aggressive incidents were recorded in these five subjects and the relationship between DRAMS assessments and the incidents was calculated. Assessments were recorded on the day of the incident, the day before the incident and on a control day, which was at least 7 days before or after an incident. Results: Four items achieved high reliability--mood, psychotic symptoms, self-regulation, compliance with routine--in addition to total score. Two items achieved intermediate reliability--antisocial behaviour and problems with/attitudes, and three items were not applicable to this client group in that none of the participants was ever rated on them--substance abuse, renewal of emotional relationships and victim access. One item had poor reliability--therapeutic alliance. There were significant differences between the DRAMS assessment on the control day and assessments both on the day prior to the incident and the day of the incident for three of the six subscales and total score. Conclusion: Initial results would suggest that the DRAMS is a reliable instrument apart from the therapeutic alliance category. Initial signs suggest that it may be predictive of aggressive incidents in residential settings. Several developments to other settings and the client groups are discussed.