A critical element of the challenges and opportunities for today's large engineering projects are associated with the multi-type and networked relationships between these projects and their various stakeholders. This dissertation advances a multidisciplinary approach-Stakeholder Value Network (SVN) analysis-as a unique lens to examine, understand, model, and manage these stakeholder relationships. The SVN approach, based on the Social Exchange Theory (SET), unifies both social and economic relationships into a common framework, under which all the stakeholder relationships are formed by the use of subjective utility analysis and the comparison of alternatives. Next, restricted and generalized exchanges are identified as two basic patterns for stakeholders to exchange both tangible and intangible value, and from this, the missing links between relationship types and exchange patterns are also discovered. In the end, the network implications, such as stakeholder importance or salience, are inferred as the outcome of both value exchanges and the structural properties of the network consisting of stakeholders and their exchange relationships. According to the above theoretically grounded assumptions, a four-step methodological framework (viz., Mapping, Quantifying, Searching, and Analyzing) is developed for the SVN analysis. As part of this development, a network utility model is built to quantify the value delivered to the focal organization (viz., the large engineering projects) through the channel of generalized exchanges. Meanwhile, the benefits from as well as a feasible way for the integration of stakeholders and strategic issues are explored under the SVN framework. In addition, for the purpose of reducing the egocentric bias associated with the pre-selection of a focal organization, the four-step framework is further developed to interpret the implications of the SVN from the perspective of the whole network. The computational challenges arising from this new development are met by the construction of a dedicated mathematical tool for the SVN analysis, namely, the Dependency Structure Matrix (DSM) modeling platform. Corresponding to the two-stage development of the methodological framework, two large real-world engineering projects are studied respectively: The first one, Project Phoenix, is a retrospective case and applies the SVN analysis from the focal organization perspective. Based on this case study, the descriptive accuracy of the SVN analysis is validated, through a comparison of important stakeholders derived from Mangers' Mental Model, the "Hub-and-Spoke" Model, and the SVN Model. Specifically, it is found that Managers' Mental Model is similar to the "Hub-and-Spoke" Model, and both models miss the Public Media and the Local Governments as important stakeholders at the beginning of the project. On the contrary, even with only prior information, the SVN Model identifies the importance of these two stakeholders by capturing the impacts of indirect stakeholder relationships as generalized exchanges. The reasons why generalized exchanges matter for today's large engineering projects are further examined from psychological, sociological, economic, and managerial aspects. The second one, China's Energy Conservation Campaign, is a prospective case and applies the SVN analysis from the whole network perspective. In this case study, five basic principles are first proposed for modeling the intraorganizational hierarchies of large and important stakeholders, and then these principles are tested as an effective means to manage the structural complexity of the SVN in the modeling process. During this process, the instrumental power of the SVN analysis is demonstrated. The SVN approach becomes complete with the above theory, methodology, tool, and meaningful findings from two representative case studies. At the end of this dissertation, two conceptual innovations are conceived to bridge the gap between the SVN analysis and systems architecting, and the theoretical, methodological, as well as empirical directions of future research on the SVN approach are also discussed.