Designing complex systems that satisfy a target reliability is difficult because of complex assembly structures and logical connections, numerous components and associated failure modes, limited reliability data or prediction models, and multi-disciplinary nature. To overcome these difficulties and to design complex systems in a systematic way, this research aims to develop a knowledge model of system design for reliability, called Reliability Object Model. This knowledge model contains a) a new failure analysis structure, b) reliability metrics that represent random failures and wearout failures, c) algorithms that allocate, predict, and assess reliability using the failure analysis structure, d) rules for design changes. The use of Reliability Object Model is demonstrated by prototype reliability tools and simplified electronic systems.