<p>Due to its practical relevance in many industrial applications, the blocking flowshop group scheduling problem (BFGSP) has gained more and more attentions in recent years. This paper presents a novel estimation of distribution algorithm-based hyper-heuristic (EDA-HH) to minimize the makespan criterion of the BFGSP. In EDA-HH, two constructive heuristics are devised based on the problem’s properties to generate high-quality initial individuals (i.e., BFGSP’s solutions). Meanwhile, an improved estimation distribution algorithm (EDA) and eleven efficient heuristics are designed in its higher and lower levels, respectively. The improved EDA is utilized not only to reasonably accumulate the valuable information of high-level permutations constructed by low-level heuristics, but also to dynamically generate excellent permutations to determine the suitable execution order of these heuristics. Moreover, to further enhance the search efficiency, two speedup scanning methods according to the problem’s properties are devised to evaluate each individual. Extensive simulations and comparisons on 560 benchmark instances demonstrate that the proposed EDA-HH can achieve better solution than nine state-of-the-art algorithms.</p>