All-atom molecular dynamics (MD) simulation has become a useful tool to investigate protein dynamics in solution or in membranes. Such simulation can provide atomic detailed information of protein dynamics on microseconds or longer timescales these days. Due to advances of simulation methodologies and computer power, MD simulation is now applied to crowded protein systems to study the relationship between the structure, dynamics, and function of proteins in the environment. Multiple proteins in the system are simulated in a solvated box at a high concentration, where proteins interact with each other via weak and non-specific molecular interactions. Protein stability and translational/rotational diffusion are compared to NMR chemical shifts and relaxation parameters, respectively. In this chapter, we review the current status of all-atom MD simulations of crowded protein systems and discuss how simulation can contribute to understanding the effects of macromolecular crowding on proteins. All-atom models of the cytoplasm in Mycoplasma genitalium were built at the atomic resolution, which includes more than a hundred million atoms and consists of proteins, RNAs, ribosomes, metabolites, ions, and water molecules. All-atom MD simulations of the cytoplasmic model were carried out using highly parallelized MD software on supercomputers and have given us several predictions about cellular crowding effects, which should be examined by future experiments.