Plants live in a microbial world and microbes have been known to influence plant health for more than a century. Remarkable progress has been made in elucidating the molecular, physiological and ecological processes in various instances of plant-microbe interactions. This has been possible thanks to a reductionist paradigm that emphasizes testing binary interactions involving only one type of microbe and one type of plant at the same time. In recent years, it has become increasingly clear that plants harbour an enourmous diversity of microbes. These observations raise important questions such as: what is the microbial diversity of the plant associated microbiota? How is the microbial diversity in the plant determined by external factors like soil biodiversity and nutritional composition? What is the role that the plant host plays in structuring the observed microbial biodiversity patterns? What are the plant genes and pathways that modulate the root microbiome and how do those interact with the environment? Finally, what is the function that the plant microbiome performs for the host? How does it influence phenotypic plasticity? and how can we manipulate the plant microbiome to modulate plant phenotypes? The work described in this dissertation provides some answers to those main questions. We characterized the bacterial diversity in and around Arabidopsis roots, and we showed that the root environment reproducibly selects for a subset of soil taxa, but we also established that the soil type is the second most important factor in explaining the observed communities inside the plant (Chapter 2). We also showed that there is weak but statistically significant effect of plant developmental stage and genotype in the root microbiome (Chapter 2). These results have been reproduced multiple times, in a variety of contexts, and represent the overarching principles of root microbiome assembly. These principles are reviewed in chapter 1 in light of current data from us and others. While natural variation revealed limited differences in root microbiome, reverse genetics approaches showed stronger effects (Chapters 4 and 5). We used mutant panels in a natural soil to find that the plant phytohormone salicylic acid, which controls a sector of plant immunity, modulates the abundance of specific taxa in the root (Chapter 4). A similar approach, found that an intact phosphate starvation response in Arabidopsis is required to assemble a wild-type root microbiome (Chapter 5). Our studies based on natural soil surveys, while useful, are limited by a lack of genomic context that is inherent to single marker surveys. To overcome this limitation, we pioneered a synthetic community approach by leveraging a large collection of wild root isolates. We have shown that we can use this approach to separate the host and environmental effects on the root microbiome (Chapter 3). We have used this synthetic community approach to delve deeper into the insights obtained in the natural soil surveys. We have shown that there is natural host genetic variation that is associated with the abundance of specific bacterial strains (Chapter 3); that plants deficient for various aspects the salicylic acid pathway can be colonized by bacteria that would be normally excluded, and that salicylic acid exerts its effect on specific strains in a direct manner (Chapter 4); finally, we have shown that a bacterial community can induce the activation of the plant phosphate starvation response, and that the master transcriptional regulator of this response is also a negative regulator of immunity (Chapter 5). Most of the synthetic community work, by us and others, is based on single synthetic communities that try to maximize diversity. While this approach has been successful, it cannot differentiate between correlation and causation, and it limits the questions that can be asked. We have developed experimental designs, and analytical pipelines that allow us to overcome these limitations. By systematically varying the microbial community composition we have shown that we can directly estimate how bacterial groups (Chapter 6) or strains (Chapter 7) will influence plant phenotypes. We can do this from a community context thereby obviating the need for binary association assays. We have shown that bacterial groups act mostly additively (Chapter 6) and that bacterial strains can act either additively or interactively depending on the plant phenotype (Chapter 7). Finally, we have shown that we can manipulate plant phenotypes by designing novel bacterial consortia (Chapters 6 and 7). Understanding plant-microbe interactions is essential for plant health and, by extension, for human health. Abating hunger is one of the great unsolved challenges of humanity. Currently, about one in nine people on Earh (∼800 million people) are hungry every day. The consequences of hunger are devastating and long-lasting. A sustainable increase in agricultural productivity is neccessary to reduce hunger and sustain projected population growth over the next century and beyond. The work described here attempts to bring together the best of reductionist and systems-level approaches, and provides key insights into plant microbiome function and manipulation that will impact conservation, mangement and agriculture.