As a first step in simulating solvent denaturation, we compare two possible potentials for urea; one based directly on a parameterization for proteins and another generated from nb initio, quantum calculations. Our results, which are derived from numerous, 1 ns simulations, indicate that both potentials reproduce essentially the same observed water structure (as evident in radial distribution functions). However, even though the quantum potential better approximates dimer energies, it is unable to simulate the dynamic behavior of water (as evident in measurements of diffusion) as well as the potential based on protein parameters. To understand its behavior in aqueous solution, we compare the urea simulations with those of solute molecules that possess the same planar, Y-shape as urea but are progressively more hydrophobic. We find that adding urea to a solution increases the number of hydrogen bonds, while adding any of the Y-shaped analogs decreases the number of hydrogen bonds. Moreover, in contrast to the Y-shaped analogs, which aggregate more as they become less polar, we find that urea mixes well in solution and has little tendency to aggregate. For our analysis of aggregation, we used a novel approach based on Voronoi polyhedra as well as the traditional method of radial distribution functions. In conclusion, we discuss how urea’s unique behavior in comparison to its Y-shaped analogs has clear implications for models of urea solvation and mechanisms of urea protein denaturation.