This paper presents the application of evolutionary programming (EP) to economic dispatch of generators having piecewise quadratic costfunctions. The proposed EP algorithm always determines the global or near global optimum without any restrictions on the shape of the costfunctions. The method is applied to an example system with piecewise costfunctions and the results are compared with those of the recent adaptive Hopfield neural network method. The test results prove that the EP method is simpler and more efficient for solving economic dispatch (ED) problems with multiple cost curves than many existing techniques.