A modified numerical algorithm is proposed to simulate cavitating flow in a convergent–divergent nozzle based on the Smoothed Particle Hydrodynamics (SPH) method. Several modifications were applied to the standard SPH equations to make it capable to model the growth, convection, and collapse of the cavitation phenomenon. As a multiphase problem with the large density ratio, some treatments were applied to satisfy the conservation of mass/momentum and guarantee the stability of the algorithm at the liquid–vapor interfaces. The pressure filter was applied to detect the cavitated particles which have pressure below the vapor pressure of the main liquid. Since SPH is a Lagrangian particle-based method, no transitional phase is defined, and sudden phase change is considered between the liquid and vapor phases. First, one of the multiphase benchmark problems was simulated to verify the functionality of the proposed SPH. Then, the simulation of cavitation growth in the convergent–divergent nozzle was investigated by comparing the results with those of commonly used grid-based methods. As a result, the modified form of the SPH method showed a desirable capability in detecting cavitated regions and modeling cavitating fluid as a multi-phase medium.