Although studied for more than a century, the spatial distribution of microorganisms in a root system still remains partly understood. In a repeated greenhouse experiment using the model plant Brachypodium distachyon, we investigated the composition and distribution of rhizosphere bacteria and their response to inoculation with artificially selected microbial communities, using two different sampling scales: root sections from distinct individual roots (apical, middle, and rear sections) and the remaining entire system recovered after homogenization. Using 16S rRNA gene sequencing, we identified that root section identity was the most influential factor on the microbiota composition (R² = 44.4%), followed by batch (R² = 34.4%), and plant identity (R² = 15.2%). Apical sections were characterized by increased abundances for Firmicutes members, while the rear sections featured more Verrucomicrobia. Root section sampling showed better sensitivity at detecting significant effects of the inoculation on the microbiota composition (e.g., local influence of inoculation on rear sections), in contrast, the homogenized sampling showed improved reproducibility (e.g., smaller sample dispersion). The comparison of the two sampling strategies highlighted a clear tradeoff between reproducibility and sensitivity, encouraging to complement traditional approaches with fine-scale sampling to improve our capacity to understand biological effects that could be otherwise missed.