In this paper, we present an innovative approach for measuring importance of project risks at a group level. First, a simulation-based risk network model is used to account for risk interactions during the project development. Various importance measures are then proposed to assess risks with respect to their influence on the global risk exposure. The originality is estimating the risk importance measures at the group level, formed using interactions-based clustering techniques. Serving as a powerful complement to the classical risk criticality analysis, the proposed method is able to support decision-making like risk response planning. An example of application to a large engineering project for implementing a tramway system is provided to illustrate the utility of the approach.