A machine’s degradation status directly influences the operational performance of the production system, such as productivity and product quality. For example, machines associated with different health states may have different remaining life before failure, thus impacting the system throughput. Therefore, it is critical to analyze the coupling between the overall system performance and the machine degradation to better production decision-making, such as maintenance and product dispatch decisions. In this paper, we propose a novel model to evaluate the production performance of a two-machine-and-one-buffer line, given the real-time machine degradation signals. Specifically, a phase-type distribution-based continuous-time Markov chain model is formulated to estimate the system throughput by utilizing the remaining life prediction from the degradation signals. A case study is provided to demonstrate the applicability and effectiveness of the proposed method.Note to Practitioners - Machine degradation is commonly observed in many industries, such as automotive, semiconductor, and food production, which gradually deteriorates the machine conditions in different operating processes and affects the production system performance. In practice, sensors are largely deployed on the factory floor to monitor the machine’s operating condition. However, a gap still exists between machine operating conditions and system performance. In this paper, we develop an analytical model to predict the machine remaining lifetime and estimate the system performance of a small scale production system, using the machine degradation signals from sensors. Furthermore, a Bayesian updating scheme is provided, which enables online evaluation by utilizing the real-time signals. Such a method provides an effective tool for production engineers to analyze the real-time system performance, and further conduct system improvements and control.