Abstract Evolutionary based Genetic Algorithms (GAs) were employed to optimize the blade shape of a Savonius-type vertical-axis wind turbine (S-VAWT) as well as the shape and the location of one doubly-curved deflector placed upstream the returning blade of an S-VAWT. To this end, GAs were incorporated into computational fluid dynamics (CFD) simulations. Significant improvement in performance was observed for the S-VAWT with optimal blades and that with the optimal deflector. The time-averaged power coefficient of the S-VAWT with optimal blades can be increased by up to 34% over that of the S-VAWT with semi-circular blades. For the S-VAWT with the optimal deflector, its time-averaged power coefficient can be boosted by 95%, compared with that without a deflector. The underlying mechanisms responsible for the performance improvement are discussed based on numerical field data from the CFD simulations.