It is widely assumed that agricultural practices have a major impact on soil living organisms. However, the impact of agricultural practices on soil microbes is poorly known, notably for species richness, evenness, and taxonomic composition. The taxonomic diversity and composition of soil indigenous microbial community can be assessed now using pyrosequencing, a high throughput sequencing technology applied directly to soil DNA. Here, we studied the effect of agriculture management on soil bacterial and fungal diversity in a tropical grassland ecosystem of northeastern Laos using 454 pyrosequencing of 16S and 18S rRNA genes. We studied soil microbial diversity of agricultural soils 3 years after conversion from native grasslands. We compared five systems: one tillage, two no-tillage rotational, one no-tillage improved pasture, and one natural grassland. Our results show first that compared to the natural grassland, tillage decreases fungal richness and diversity by ?40 % and ?19 %, respectively and increases bacterial richness and diversity by +46 % and +13 %, respectively. This finding evidences an early impact of agricultural management on soil microbial diversity. Such an impact fits with the ecological concept of "intermediate perturbation"-the hump-backed model-leading to classify agricultural practices according to the level of environmental stress they generate. We found also that land use modified soil microbial taxonomic composition. Compared to the natural pasture, tillage decreased notably the relative abundance of Actinobacteria (by ?6 %), Acidobacteria (by ?3 %) and Delta-proteobacteria (by ?4 %) phyla, and by contrast increased the relative abundance of Firmicutes (by +6 %), Gamma-proteobacteria (by +11 %), and Chytridiomycota (+2 %) phyla. We conclude that soil microbial diversity can be modified and improved by selecting suitable agricultural practices. Moreover no-till systems represented intermediate situations between tillage and the natural pastu