Studies of social information use have generally adopted an objectivist definition of 'information', treating it as an entity which resides in documents such as social enquiry reports and whose effect can then be measured as the corr7latio~ of inpu~ (information) with output (decision). Cons~derat1on of mag1strates' perceptions has been partial and problematic. The present study seeks to effect two major, interrelated, shifts in the study of social information use. Firstly, utilising the sociologies of knowledge and science, information is redefined as a product of the active creation of knowledge representations from data by decision makers. The creation of representations is seen to occur according to conventions of interpretation, generated as decision-makers seek to render their everyday activities coherent and meaningful, and acting as a cultural resource to assist in the accomplishment of future practices. However, information-creation is never neutral. In relating to practices it embraces the character of social relations and the assymetries of power inhering in these. A 'knowledge/power' analysis is adopted which enables 'information' to be viewed in relation to the micro-processes of organizational arenas and to social relations across time and space. Secondly,this forms the context for an empirical study of the generation and deployment of social information-as-representations in the juvenile court. Magistrates were interviewed and observation undertaken in six juvenile courts. The focus is on the decoding of social data by magistrates, both from social enquiry reports and other sources (solicitors, parents, defendants themselves). In decoding social data magistrates utilise conventions of interpretation which are dominated by a search for disciplinary control indicators. Reports are seen as malleable resources whose use is determined more by the decoding context into which they are sent than by their intrinsic properties. Control indicators are manufactured from social data to render the business of tariff sentencing possible and meaningful; the deployment of social information is a fulcrum of the classification of offenders along the 'slippery slope' of bifurcatory sentencing. Magistrates' accounts are thus situated in relation to the practices of the court and interorganizational boundaries, and ultimately are related to 'long distance' control. The 'social enquiry' is found to be, not a narrowly mundane matter of providing 'information for the court', nor solely a locally exerted power, but a far reaching technique of power which must be situated in relation to concepts of the 'tutelary complex' and the 'carceral continuum'.