Effective risk management is a central function in the successful planning and execution of large infrastructure projects. This paper explores how current knowledge-based approaches for risk management can be improved upon so that they are more responsive to the attributes of a project and the needs of system users. A review of existing knowledge-based systems for risk management provides a backdrop for a discussion on desirable characteristics of such an approach. The proposed methodology adopts a model-based technique in that explicit abstractions of project components and processes, and the physical, regulatory, political, social, financial, economic, contractual, and organizational environments in which they are located, are created to assist in the reasoning about possible risks. This contrasts with several current systems that use only implicit representations. The reasoning schema and models of the physical project and environment that are used for the reasoning process are described in the paper.Key words: risk identification, project modeling, knowledge management, infrastructure projects.