Expert systems have during the last fifteen years successfully been applied to a number of difficult problems in a variety of different application domains. Still, the impact on the commercial market has been less than expected, and the predicted boom just failed to occur. This thesis seeks to explain these failures in terms of a discrepancy between the tasks expert systems have been intended for and the kind of situations where they typically have been used. Our studies indicate that the established expert systems technology primarily focuses on providing expert-level solutions to comparatively well-defined problems, while most real-life applications confront a decision maker with much more ill-defined situations where the form of the argumentation rather than the explicit decision proposal is crucial. Based on several commercial case-studies performed over a 10-year period together with a review of relevant current research in decision making theory, this thesis discusses the differences between different use situations with respect to the degree of how well-defined the decision task is and what kind of support the users require. Based on this analysis, we show the need for a shift in research focus from autonomous problem solvers to cooperative advice-giving systems intended to support joint human-computer decision making. The requirements on techniques suitable to support this trend toward cooperative systems are discussed and a tentative system architecture and knowledge representation for such systems is proposed. The thesis concludes with a research agenda for examining the cost and benefits of the suggested approach as a tool for cooperative advice-giving systems, and to determine the appropriateness of such systems for real-world application problems.