Complex electromechanical systems considering performance degradation are multi-state systems which cannot be described by traditional reliability theories based on the binary states assumption. Generally, such systems contain both electronic and mechanical components that have different failure mechanism and characteristics. In addition, states of the system and its components are time-variant, which makes reliability modeling and analysis challenging. To overcome these difficulties, a hierarchical modeling approach is proposed in this paper. At the component level, we adopt a homogeneous Markov chain model for electronic components and a multi-state physical model based on Monte Carlo simulation for mechanical components respectively. After that, a time-variant multi-state multi-valued decision diagram (TMMDD) method is proposed to represent the system structure function at the system level. Finally, a practical engineering example of a space drive system is provided to illustrate the proposed method.