Abstract The authors propose an efficient algorithm, NB-MAFIA, for mining maximal frequent itemset using NList, which uses node list of prefix tree to represent itemsets. By using N-List, itemsets’ support can be efficiently computed because of the high compactness of N-List and the efficiency of the method to intersect two N-Lists. Meanwhile, the authors employ some search space pruning strategies and superset checking strategy to improve NB-MAFIA. To evaluate NB-MAFIA, the authors compare proposed algorithm with two state-of-the-art algorithms on a variety of real and synthesis datasets. Experimental results show that NB-MAFIA is efficient and outperform the baseline algorithms in most case. Especially, NB-MAFIA is more efficient on dense datasets.