The trend toward microelectronics and higher system sophistication requires the development of more advanced simulation and performance prediction techniques. In most large electronic systems the bulk of the total circuitry (60–80 per cent) is made up of multiple usage of relatively few basic standard circuits. The remainder is generally composed of special circuitry unique to that particular system's characteristics. In view of the large quantities of the basic circuits which are used over and over again, it becomes important from the standpoint of reliability, specification setting, and production control to subject these circuits to the most stringent evaluation possible during the design phase. This can be accomplished by supplementing the test and evaluation phase with extensive computer simulation and performance prediction effort. Fortunately, modern high-speed digital computers permit this to be done in an economic manner. North American/Autonetics has designed a number of special computer simulation and performance prediction programmes. This article will discuss two of the more frequently used programmes; namely, a programme implementing the Worst-Case method and another which implements the Monte Carlo method. In addition, a third approach will be described which combines both methods. These programmes require mathematical models for simulation of the electronic components. The models, primarily the diode and transistor mathematical models, will be presented in detail. Special emphasis is given to the non-linear model characteristics required to closely duplicate physical behaviour.