It is well-known that elliptical-bladed Savonius wind turbine yields relatively better performance than conventionally used semicircular-bladed turbines. This is mainly due to lesser tip loss and delayed flow separation that allow the elliptical turbine to acquire higher rotational speeds than semicircular turbine under a given wind load. In this work, an experimentally-validated inverse analysis is done to determine the optimum blade configurations involving the chord length, turbine height, aspect ratio, and the necessary overlap ratio to derive a required power and torque from elliptical-bladed Savonius wind turbines. Due to obvious advantages of evolutionary metaheuristic optimizers in general, here differential evolution (DE) search algorithm is used to solve the inverse problem through a least-squares minimization of the relevant objective function. The objective function is further subjected to feasible bounds of the unknown design variables. The effects of blockage corrections are duly considered and the variations of the design variables along with the objective function are studied over a range of iterations of the DE algorithm. Through comprehensive analysis, it is highlighted from the present study that for harvesting a given performance, rotor swept area can be reduced by 6.25% with respect to the available experimental data under identical operating conditions of the wind turbine. Multiple blade configurations can be acquired, all of which invariably satisfy the required performance criterion. This study also highlights that amongst various dimensional parameters, turbine height and aspect ratio play more prominent role than chord length and overlap ratio and the blade chord influences only the torque but not markedly the power. The results obtained from the present work are proposed to facilitate the concerned designer to explore various feasible blade designs and determine the suitable one, thereby avoiding valuable time elapsed in repetitive fabrication and testing of various designs.