Pumped storage is generally viewed as the most promising technology to increase renewable energy source (RES) penetration levels in power systems and particularly in small autonomous island grids. Combined wind and pumped-storage "virtual power plants", called hybrid power stations (HPS), constitute a realistic and feasible option to achieve high penetrations, provided that their components are properly sized. In this paper, the optimum sizing is investigated for a pumped storage HPS operating in an island system. The analysis addresses the sizing of the main HPS components (hydro turbines, pumps, wind farm, reservoirs), adopting either the investor's perspective, where the objective is to maximize the return on the HPS investment, or a system perspective, where the optimization target is the maximization of RES penetration, along with maintaining the lowest possible generation cost in the system. Genetic Algorithms (GAs) are applied for the optimization and a real isolated island power system is used as a study case. The adopted operating policy and pricing principles, which critically affect the optimal sizing of an HPS project, are based on the existing regulatory framework for storage stations in Greek islands.