The reliability of a system is usually a function of several variables and parameters, most of which are random variables. In the case of large complex systems, collection of data sufficient to give a complete characterization of performance may not be possible. In general it is quite difficult to find the density function for a function of random variables. Knowledge of the moments of the transformed variable can be quite useful. In probabilistic design, the design variables and parameters are identified explicitly, and stress and strength distributions determined. Probabilistic design has its origins in the aerospace industry. Basic steps in the process include: formulation of an initial design; estimation of external forces and their probability density functions; and analysis of the initial system, and design criteria. Reliability can be extremely sensitive to input errors. The distribution of stress and strength variables may also be functions of time.