Abstract: In response to the current situation of static fragmentation in tunnel construction safety assessment systems, insufficient multi-source information, and the challenge of ensuring the reliability of assessment methods, this paper proposes an integrated dynamic safety assessment technology and engineering application system for tunnel construction based on multi-source information fusion. With a multi-dimensional and multi-scale concept, the tunnel construction cycle is divided into three nested scales: survey and design, advance prediction, and excavation monitoring, as well as three source dimensions under each scale: geological information, construction information, and prediction information, to form a multi-source safety information assessment index system. This system can meet the requirements of survey and design in alignment with construction progress, two overall risk assessments under the advance prediction scale, and various specialized construction safety assessments under the excavation monitoring scale. To enhance the rationality and reliability of safety assessment results, the D-S evidence theory serves as the fusion framework. The interval Euclidean distance method and average evidence method optimize the basic probability assignment calculation and conflicting evidence fusion in the multi-source information fusion process. The implementation of this comprehensive dynamic safety assessment method and system is illustrated through a real tunnel engineering project. Results show consistency between the assessment outcomes and actual on-site conditions. Further discussion, comparison, and verification demonstrate its rationality and reliability, providing valuable reference and practical guidance for tunnel construction safety assessment and control.