In reliability-based design optimization (RBDO) for the satellite system, reliability modeling is a very important issue. However, due to the limited failure information and small scale production in space engineering, it remains challenging as how to properly construct the reliability model especially in the early conceptual design phase for a new satellite. Actually, there is multi-source lifetime (or failure) information of satellite subsystems or systems from the past missions or designer experiences. How to fully integrate the multi-source information so as to enhance the reliability model with more knowledge is an open problem to be solved. To address this problem, in this paper, it is proposed to use the Bayesian Melding Method (BMM) to merge the multi-source lifetime information from the multi-level structure of satellite system and subsystems, based on which the lifetime distributions can be updated and the satellite reliability model can be properly constructed. With the enhanced reliability model, the satellite RBDO is further studied to optimize the satellite design with budget constraints on mass, cost, and reliability. Finally, two case studies are utilized to demonstrate and verify the proposed method. INDEX TERMS Satellite system design, reliability-based design optimization, Bayesian melding method, multi-source lifetime information.