Within the electric power industry, the average company's enterprise system - i.e. the overall system of IT related entities - is today highly complex. Technically, large organizations posses hundreds or thousands of extensively interconnected and heterogeneous single IT systems performing tasks that varies from enterprise resource planning to real-time control and monitoring of industrial processes. Moreover are these systems storing a wide variety of sometimes redundant data, and typically they are deployed on several different platforms. IT does, however, not execute in splendid isolation. Organizationally, the enterprise system embraces business processes and business units using as well as maintaining and acquiring the IT systems. The interplay between the organization and the IT systems are further determined by for instance business goals, ownership and governance structures, strategies, individual system users, documentation, and cost.

Lately, Enterprise Architecture (EA) has evolved with the mission to take a holistic approach to managing the above depicted enterprise system. The discipline's presumption is that architectural models are the key to succeed in understanding and administrating enterprise systems. Compared to many other engineering disciplines, EA is quite immature in many respects. This thesis identifies and elaborates on some important aspects that to date have been overlooked to a large extent. Firstly, the lack of explicit purpose for architectural models is identified. The thesis argues that the concerns of a company's Chief Information Officer (CIO) should guide the rationale behind the development of EA models. In particular, distribution of IT related information and knowledge throughout the organization is emphasized as an important concern uncared for. Secondly, the lack of architectural theory is recognized. The thesis provides examples of how theory, or analysis procedures, could be incorporated into the Enterprise Architecture approach and hereby concretely drive the development of the architectural models. Due to the nature of enterprise systems, EA theories inevitable will be of an indicative character. Finally, in relation to the models as such, three aspects are highlighted. Firstly, the cost of collecting information from the organization to populate models is routinely neglected by the EA community. This expense should be evaluated in relation to the utility of analyses that the information can provide in terms of better informed decision making by the CIO. Secondly, models (and meta-models) must be kept consistent. And thirdly, the design of models is restricted by the limited mental capabilities of the minds of the model users. CIO concerns must consequently be easy to extract from the Enterprise Architecture models.

Key words: Enterprise Architecture, Enterprise System, Chief Information Officer (CIO), Information Technology (IT) Management, Architectural Theory, Electric Power Industry