The amount of data available on the microbial world rapidly increases, but basic microbial ecology does not benefit from these data at their full potential because of a lack of suitable modelling approaches. Accordingly, my research consists in developing probabilistic models to explore and summarize data structure and compare it to theoretical expectations, so as to advance our understanding of microbial community assembly, dynamics and function. To do this, I combine mechanistic models developed in classical ecology and evolution with modelling approaches that are tailored to microbial data. I apply these models to diverse ecosystems spanning host-associated microbiota, soil microorganisms, and ocean plankton.