To achieve accurate and effective safety risk assessment for shield tunneling underneath existing tunnel, a set of safety risk assessment methods based on the fuzzy Bayesian network and evidence theory is proposed. First, by summarizing a large amount of literature and relevant standards, a set of relatively perfect safety risk assessment systems and evaluation standards for shield tunneling underneath existing tunnel is established by selecting 14 main influencing factors. Second, the evidence theory that can effectively fuse uncertain information is introduced, and the safety evaluation model of shield undercrossing existing tunnel is established by combining it with fuzzy Bayesian network. Third, a metro project in Wuhan is taken as an example to conduct an empirical analysis. Conclusions are drawn as follows: (1) based on the risk reasoning and sensitivity analysis of the constructed model, the safety risk state of the project is determined to be general and tends to be more dangerous; and (2) the key control factors are the clear spacing of the tunnel, diameter of the new tunnel, plane angle of the two tunnels, internal friction angle, complexity of the construction environment, and complexity of the construction coordination.