Cells assemble large, non-membrane bound granules of protein and RNA, termed Ri- bonucleoprotein granules (RNP granules), often in response to a wide variety of cellular stresses. This behavior is conserved from yeast to mammals. Some RNP granules ap- pear important in the stress response, while others are important for proper organismal development, and still others for control of RNA degradation and transport. Curiously, proteins found within granules are disproportionately likey to contain Intrinsically Dis- ordered Regions. Here, I show that those disordered regions can often drive higher order assembly in vitro and contribute to granule assembly in vivo. I found that these domains can make it easier for proteins to undergo a process known as Liquid-Liquid Phase Separa- tion in response to changes in ionic strength, wherein the protein of interest self-partitions into a concentrated liquid phase. The droplets that form mimic many of the behaviors of RNP granules in cells, such as recruitment of other IDR-containing proteins, assembly in response to RNA, and rapid exchange of contents with the surrounding medium. I also found that proteins that form these droplets tend to aggregate over time, turning the dynamic droplets into static structures. Further, I identified several limitations to my in vitro model, most importantly the impairment of IDR-based phase separation in the presence of other proteins or cellular lysates. However, I also helped uncover the synergistic relationship between IDRs and the more well studied protein-protein and protein-RNA interactions that are important for granule assembly. I therefore propose an inclusive model of granule assembly which asserts that a wide variety of types of interactions are important, and that it is the sum-total of these interactions that determines whether or not a granule assembles. Advisors/Committee Members: Roy Parker, Christopher Link, Loren Hough, Marcelo Sousa, Amy Palmer.