Sensitivity analysis is an important adjunct of any mathematical optimization problem. In this paper, the various types of sensitivity relations are explained and interpreted in the context of design optimization of complex hardware systems, and the practical significance of sensitivity theory for the design engineer is illustrated. Computer simulations have already become conventional in engineering design. Direct optimization procedures, developed recently, employ mathematical programming techniques with the aim of finding the "best" design parameters directly rather than by trial and error. These optimization techniques, despite their great utility, do not provide the engineer with all the answers he seeks; they do not answer such questions as: "How dependent is the optimum design on certain technical, cost or marketing assumptions?", "By how much should the design parameters be changed when these assumptions are modified?", "How flat is the objective function around the optimum?", and "Was the selected objective meaningful in the first place?" An elegant and often easily obtainable answer to these questions is provided by the sensitivity methods developed and illustrated here.