Previous research has indicated that operators of complex systems will achieve higher system operation efficiency and better job performance when they hold high quality mental models. Authors of recent studies have called for greater attention toward how to improve the development of mental models especially intervention factors influencing mental model quality. The current study aimed to develop a new training strategy called mental model-based training and to examine its effectiveness by conducting a quasi-experimental study. Results demonstrated that operators exposed to mental model-based training showed more frequent reflective and learning behaviors in later work than those receiving no training, whereas mental model-based training did not result in superior mental model quality over the control condition. Findings were discussed in terms of the advantages of the mental model-based training strategy as well as its applications in complex systems.