Cyclic variation is a serious problem for two stroke engines at low load conditions. The occurrence of variations seems randomly. However, there are some rough rules behind this chaotic process. The combustion of one cycle is affected by the concentration of residual gas in cylinder, which is the results of the scavenging process of pre-cycles. The objective of this paper is to study how variations occur in consecutive cycles with conventional engine cycle simulation analysis, i.e. a deterministic approach was tried to simulate a random process.The cycle model of a crankcase scavenged type single cylinder two-stroke engine was developed in this paper. The engine was divided into four sub-systems, each one is an independent thermodynamic system. Two zone combustion model was used and two different scavenging process, the perfect mixing model and the perfect displacement model, were considered in this paper. Results of simulation show that al some running conditions, the engine runs at a steady state, i.e.. the engine performance converges in a one-cycle period. However, at some other conditions, the engine performance might converge in a two-cycle period or even multi-cycle periods. For some special cases, the engine performance would never converge, and random variations of cycle performance were observed.Some factors that may affect the cycle behavior were investigated in this study. Results of simulation show that the engine load is the most important factor to determine the cyclic variations. Other factors, such as flame speed, heat transfer, misfire condition, also have some degrees of influence on cyclic variations. Perfect mixing scavenging model may, result in highly varied cycles. However, perfect displacement scavenging model would produce cycles with no variations.