In this paper, we develop a resilience modelling and analysis framework for multistate systems. The developed framework is based on Markov reward processes, in which the performance degradation and recovery processes of the multistate system are modeled by a continuous time discrete state Markov chain, and the losses caused by disruptive events are modeled by reward rates. A numerical metric is defined to quantify resilience based on the losses the system suffered from the disruptive events. The developed methods are applied to analyze the resilience of a nuclear power plant under the threat of earthquakes.