The total sputtering yields for water ice due to kiloelectronvolt cluster bombardment have been measured and compared to the predictions made by the mesoscale energy deposition footprint (MEDF) model. For C60 bombardment, the experimental yield varies almost linearly from 820 water molecule equivalents at an incident kinetic energy of 10 keV to 10 100 water molecule equivalents at a kinetic energy of 120 keV. For Au3 bombardment, the experimental yield varies almost linearly from 630 water molecule equivalents at an incident energy of 10 keV and rises to 1200 water molecule equivalents at 25 keV. The MEDF model is used to calculate relative yield trends with respect to incident energy using short-time molecular dynamics simulations. The results of these calculations indicate that the model can effectively predict the yield trends observed for these two clusters in experiments, although there is a consistent overestimate of the predicted induced C60 yield. It is hypothesized that this overestimate can be explained by the absence of reactions and ionization processes in the current simulations. Despite this omission, experimental yield trends can be accurately predicted using relatively small amounts of computer time. The success of the model in predicting the yield of water from ice films using a variety of energies and projectiles suggests this approach may greatly aid in the optimization of experimental configurations.