Abstract : Workload smoothing on assembly lines, which aims to evenly assign tasks to stations, supports workforce planning and resource optimization. In this paper, we study smoothing assembly lines and develop a problem-specific heuristic to efficiently solve large-sized instances. To build solutions, the algorithm uses a number of well-known priority rules for task assignment in conjunction with a probabilistic decision-making procedure for closing workstations. We conduct an experimental design for selecting the best performing priority rules and for tuning the probabilistic decision-making procedure. The efficiency of our algorithm is tested and demonstrated through an extensive experimental study.