In-cylinder pressure traces vary significantly from cycle-to-cycle in spark-ignition (SI) engines. The variations, substantially present even when engine is stable, are magnified under certain engine operating conditions. As a result, engine torque output oscillates and engine operation becomes unstable. EGR tolerance, lean burn limit and spark retard capabilities at CSSRE (Cold Start Spark Retard and Enleanment) are mostly determined by the levels of cycle-to-cycle variations. None of the engine computer models, however, have included cyclic variations for routine industrial applications. As the application domain of engine simulation models expands into unstable engine operating conditions, the modeling of cyclic variations becomes increasingly important. In this research, reviews were conducted regarding different approaches for the simulation of cyclic variation. Discussions and model results have shown that the effective and practical way to simulate cycle-to-cycle variations was by using nonlinear regressions. This paper documented the development and preliminary validation of such a regressed computer model in an engine cycle simulation model.