This thesis investigates whether a game world based solely on fluid dynamics could be an interesting alternative to one based on rigid bodies by presenting a prototype to players where they can control the movement of a fluid. The players evaluate the prototype’s responsiveness, intuitiveness, visual appeal, immersion, and how well they think the fluid could represent ghosts, flocks, magic, and a platformer avatar by rating them vocally while playing. The fluid physics engine prototype is a simple grid-based one that utilizes the ideal gas law and pressure gradient force to calculate the flow between its cells. It is unstable and thus cannot handle too large time-steps. Therefore, simulating many cells (more than 10000 for two fluids) at the same is a slow process, and the prototype can only present a pixelated result. If a highresolution game is in general preferred by players, a computationally efficient solution could consist of utilizing shaders for cell blending. In the experiments, all categories except the platform avatar received positive evaluations despite the prototype’s early stage, which according to theories presented within Swink’s Game Feel (Swink, 2009) and the MDA framework (Hunicke et al., 2004) may indicate that a game based on a fluid physics engine like this one could be entertaining. The simulation could possibly be run as a separate system to add game mechanics to a game based on rigid bodies as well.