Abstract: To help control the investment risks regarding three commonly over-budget items in river-crossing tunnels, a risk-identification model based on grey fuzzy multi-attribute group decision making was established with respect to fuzzy, grey-level, multi-attribute and group risk characteristics. The model's feasibility and effectiveness were verified by using the Nanjing Weisan Road river-crossing project as an example. The results show that it is possible for the model to achieve risk sequencing by making use of expert opinions in a comprehensive and realistic way, helping decision makers to identify the key, major and general potential risks more accurately.