Artificial selection applied at community level is an important, but still growing topic in the field of ecology and experimental evolution [1-3]. Its recent implementation to microbial communities holds not only appealing promises in terms of fundamental knowledge about selection itself [3], but also in terms of relevant applications to our society, including bioremediation [4] and plant traits enhancement [5]. Here we transposed the concept of artificial selection of communities to perform experimental evolution of root microbiota inducing relevant phenotypic changes in plants. We grew ten successive generations of four weeks old Brachypodium distachyon inoculated with artificially selected root microbiota from the previous generation, corresponding to ~3700 plants. Depending on experiment goals, selection was applied based on specific plant phenotypic traits of interest such as aboveground biomass or leaves color nuances as a proxy of nitrogen content, using an automated high-throughput plant phenotyping platform. We orientated evolution in different directions by respectively selecting plants within several lineages displaying the lowest and the highest values for targeted traits against random selection controls. Root microbiota were characterized during the selection experiment by means of 16S rRNA gene amplicon sequencing. Despite challenges associated with ensuring efficient heredity of selected communities and plant phenotypic changes, results obtained showed rapid response of plants to artificial root microbiota selection, with significant divergence for targeted traits after few generations. Our results support the fundamental notion that plant phenotypic changes may be rapidly acquired via artificial selection of root microbiota, which could potentially contribute to rethink the way we select for desired plant functions.