Setting kanban sizes is one of the first decisions that users of kanban system must address, yet researchers have largely assumed kanban sizes to be given. This paper investigates the effect of varying kanban size on the performance of just-in-time (JIT) manufacturing systems. Two types of JIT production systems, the Pull-type and the Hybrid-type are analysed using computer simulation models. The performance measures considered simultaneously are the fill rate, in-process inventory, and manufacturing lead time. Parameters such as demand rate, processing time, and kanban size are taken into consideration, thereby finding the possible solutions of the kanban size that can be employed to achieve the most favourable conditions for production. A favourable condition usually refers to the ability of the system to produce finished goods at a shortest possible lead time, which the customers are always demanding for. Both the single product and multi-products manufacturing environments are investigated.