Predicting how microbial communities respond to environmental change is crucial for managing their function in the face of changing climate, health, and disease. Yet these communities’ enormous diversity and complexity raises the question of whether it is possible to predict their dynamics. While the environment plays a direct role in selecting for particular species, measuring these direct effects may not be sufficient for making predictions, because the environment also indirectly alters interactions between species. Moreover, measurements of communities in particular conditions may not predict their behavior when the environment fluctuates between conditions. In this talk, I will explain how I have employed tractable laboratory microcosms and mathematical models to formulate unifying rules of microbial communities subject to changing temperature, mortality, and fluctuations. I will also outline future plans to study longer-term community dynamics and to predict when evolution will maintain or destabilize coexistence.