[Background] Soil sampling is the basis of soil research, and different sampling designs can have a considerable impact on the results of soil microbial research. [Objective] To investigate the effects of different soil sampling designs on microbial community structure and diversity of soils obtained through 16S rRNA gene high-throughput sequencing. [Methods] In this study, soils from two different habitats were sampled by grid sampling method, and all the 18 soil samples were analyzed by 16S rRNA gene high-throughput sequencing. Five common soil sampling methods were simulated through combining different points from the sampling grids. The sequencing results of different sampling methods were simultaneously compared. [Results] Different sampling methods resulted in different soil microbial sequencing results. The number of bacterial species gradually increased with the increased of numbers of samples, and the growth rate would be gentle when numbers of samples were greater than 5; dominant species (more than 200 sequences) could be wholly observed in few numbers of samples (1?3).; the Shannon-Wiener index and Simpson index were similar, when the numbers of samples was from 1 to 3, both indices increased significantly, then it slowed down. [Conclusion] In the study of soil bacterial microbial sequencing, numbers of soil samples below 3 would affect the reliability of sequencing results. Comparison of the sampling methods, Quincunx sampling or serpentine sampling method is more suitable for soil sampling design.