The development and dissemination of Electronic Medical Records (EMR) is impeded because of several reasons. One of the reasons is considered the 'unruliness' of narrative data. In this article it is illustrated how the medical record can be structured to make it accessible from different perspectives, without the need to change the free text format of the narratives. The organising principles behind three existing medical record structures are analysed: the source-oriented, problem-oriented and time-oriented medical record. These principles are combined in a model of the medical record with four different views on medical data: a typological view, a time view, a problem view, and a process view. A data model is presented in which each of these views can be defined. In the discussion some examples are given of EMRs in which some of these views have been applied on coded data.