Abstract: With the development in electricity system to an integrated energy system, the traditional energy management system is also expected to be developed and applied to the integrated energy system. As the core of the energy management system, state estimation is an essential part of analyzing the operation state of system, and is the basis for realizing functions such as optimal scheduling. Therefore, a two-level robust state estimation method for the integrated electricity-heat system was proposed. In the first-level state estimation, the robust state estimation model of the electricity and hydraulic networks based on the second-order cone constraints was constructed by introducing auxiliary variables. In the second-level state estimation, the estimated measurement values obtained by the first-level state estimation were used to establish a robust state estimation model of the thermal network after the nodes' temperature constraints are taken into consideration. The case study consists of the 32-node heat system in Barry Island and the IEEE 14-node power system. The results show that the proposed method has good estimation accuracy and robustness, and the two-level estimation strategy significantly improves the estimation in the thermal network.