Soil microorganisms such as mycorrhizae and plant-growth-promoting rhizobacteria have beneficial effects on crop productivity. Agricultural practices are known to impact soil microbial communities, but past studies examining this impact have focused mostly on one or two taxonomic levels, such as phylum and class, thus missing potentially relevant information from lower levels. Therefore, we propose here an original, sub-phylum method for studying how agricultural practices modify microbial communities. This method involves exploiting the available sequence information at the lowest taxonomic level attainable for each operational taxonomic unit. In order to validate this novel method, we assessed microbial community composition using 454 pyrosequencing of 16S and 28S rRNA genes, and then we compared the results with results of a phylum-level analysis. Agricultural practices included conventional tillage, reduced tillage, residue removal, and residue retention. Results show that, at the lowest taxonomic level attainable, tillage is the main factor influencing both bacterial community composition, accounting for 13 % of the variation, and fungal community composition, accounting for 18 % of the variation. On the other hand, phylum-level analysis failed to reveal any effect of soil practice on bacterial community composition and missed the fact that different members of the same phylum responded differently to tillage practice. For instance, the fungal phylum Chytridiomycota showed no impact of soil treatment, while sub-phylum-level analysis revealed an impact of tillage practice on the Chytridiomycota sub-groups Gibberella, which includes a notorious wheat pathogen, and Trichocomaceae. This clearly demonstrates the necessity of exploiting the information obtainable at sub-phylum level when assessing the effects of agricultural practice on microbial communities.