An issue in the development of large and complex systems is that the design process is essentially open-ended. The systems engineering phase finds an acceptable theoretical design and passes it on for manufacturing. Little attention is paid to "real-world" detrimental effects of manufacturing and operational variabilities upon the "designed" system performance levels. Alternatively to try to account for search effects, a "worst-case" system design is often attempted with predictable impact to system costs. As a result of the above issues, the importance of running statistically based system and subsystem level simulations has become particularly critical in this period of increasingly complex commercial systems, defense conversions and dual-use technologies. Monte Carlo techniques could be considered but often lead to excessively time-consuming and inefficient design tools. We present a means of coupling more efficient statistical techniques (based on a design of experiments methodology) to conventional systems simulators. The resulting software tool, called STADIUM, is currently in use within industry to characterize system performances in terms of circuit, subsystem and system level variabilities.