large-scale bridges in China built in the 1990s has entered a peak maintenance period, with frequent occurrences of disasters and accidents, such as scour, collapse, and fire. However, an efficient and accurate risk assessment methodology for targeted risk management of large-scale bridge infrastructure is lacking. To address this, a semi-quantitative safety risk assessment methodology for in-service bridge operations under multiple hazards is proposed, based on statistical theory and a large sample size. First, a database of bridge safety incidents in China (591 incidents) was established, along with the collection of over 1 million related incident statistics. A bridge risk probability model is developed using statistical theory and large sample techniques, and a scalable risk assessment index system is created based on the hierarchy of risk sources. Within this framework, risk assessment methods for various types of disasters can be developed according to data characteristics. The application of this generalized method is demonstrated through a case study of fire risk assessment.