Abstract: Multiple sequence alignment plays an important role in sequence analysis, including identification of functionally important sites and phylogenetic analysis. At present, most alignment software uses the strategy of progressive alignment or iterative alignment, but both strategies have high time complexity, so it is difficult to deal with the alignment problem of long sequence and large datasets. Although star alignment has a very low time complexity, the accuracy of star alignment is not ideal, so it only applies to sequences with very high similarity. To solve this problem, we introduce profile alignment in progressive alignment to improve the accuracy of star alignment algorithm and avoid significantly increasing the time complexity of star alignment. Experiments show that the improved star alignment algorithm can effectively improve the accuracy of the alignment.