A genetic algorithm (GA) approach to reconstruction of the locations and shapes of two-dimensional impenetrable objects from scattered field data is presented. Local shape function method is used in the forward model, while a GA is used to seek for a solution minimizing the discrepancy between the measured and the estimated scattered fields in the inverse process. Some GA operators appropriate for the 2-D objects are proposed to achieve a high convergence rate. Several numerical results are presented to demonstrate the effectiveness of the GA approach.