A shape optimization problem is increasingly popular in various parts of industry and academia. Several shape optimization approaches emerged in the past few years. Population-based methods (PPM) are among them, with convenient attributes and characteristics that lead to a discovery of proper solutions of investigated problems. Crucial quality of PPM lies primarily in number of randomly scattered individuals, which can move through a whole given computational area. The chosen PPM in this research of a finding of a proper design of the centrifugal pump impeller is Particle Swarm Optimization Algorithm (shortly PSOA). This algorithm is strongly influenced by a social behaviour of various animals, such as ants or fish. Each member of the swarm moves in the whole given computational space and is strongly attracted to an individual with the best value of the examined function, e.g. pump efficiency. The research focuses on a new tool for automatic shape optimization based on the Multi-objective PSO and its outcome - three optimized designs of the centrifugal pump impeller. These three impellers are compared on the performance characteristics basis. The essential differences are discussed, and the design of a suitable impeller is outlined.