AbstractThe Taguchi idea of parameter design has received considerable attention from statisticians and engineers. However, it is only recently that some of the underlying principles of his ideas have been clarified and other alternative procedures have been proposed. This paper reviews the currently available procedures for parameter design in terms of minimizing expected loss. Advantages and disadvantages of each procedure are discussed for the three types of performance characteristics, and guidelines are provided for the choice of an appropriate procedure for a given parameter design problem.