To identify ways to improve the predictive capability of the current RANS-based cavitating turbulent closure, a filter-based model (FBM) is introduced by considering sub-filter stresses.The sub-filter stress is constructed directly by using the filter size and the conventional turbulence closure.The model is evaluated in steady cavitating flow over a blunt body revolution and unsteady cavitating flow around a Clark-Y hydrofoil respectively.Compared with the experimental data, those results indicate that FBM can be used to improve the predictive capability considerably.