Abstract. The health and performance of all macroorganisms depends on complex communities of microorganisms − bacteria, viruses and fungi that interact with each other and their hosts. Host-associated microbial research has accelerated in the last decade, resulting in advances in our understanding of the structure and taxonomic composition of these ‘microbiomes.’ Yet, microbiome dynamics remain difficult to predict, and the mechanisms underlying the collapse and reassembly of disturbed host-associated microbial ecosystems are largely unexplored. This is in part because descriptive empirical studies have outpaced the development of mathematical models in microbiology, and emphasis in microbiome studies to date has been on cataloguing variation in species diversity and taxonomy across habitats. A general and predictive theory of microbiome disturbance would bolster our understanding of relationships between host health and symbiont dynamics, and inform agricultural, environmental and biomedical efforts to engineer microbiomes that benefit society. This working group will bring together experts in microbiology, computational biology and theoretical ecology to develop new mechanistic models that predict microbiome response to disturbance, and to test these models with existing data.