Remanufacturing has in recent decades become an alternative way for sustainable development to cope with the increasingly resource crisis and environmental pollution problems. The line balancing of remanufacturing is more complex than that of traditional manufacturing due to the uncertainty in the remanufacturing process, especially as the assembly time is uncertain. As a result, the line balancing of the remanufacturing assembly has become a critical issue. This paper evaluates the uncertain assembly time in each station based on an approach of Fuzzy Graphical Evaluation and Review Technology (FGERT) network. A remanufacturing multi-objective mixed-flow assembly model is established by optimizing the cycle time, assembly line smoothing coefficient and balance time loss rate. Then an adaptive bilayer genetic algorithm is proposed to solve the optimization problem. Finally, an example was given to demonstrate the effectiveness of the proposed method. Results show that the production rhythm balance of remanufacturing assembly is significantly improved, and the idle time in each assembly station has been reduced as well. The cycle time of the remanufacturing has reduced from 344s to 336s, with the smoothing coefficient and balance time loss rate of being 8.73 and 1.69% compared to that without assembly line balancing. The proposed method provides a useful tool for improving mixed-flow remanufacturing assembly lines in an electromechanics industry. Copyright (C) 2022 The Authors.