Competing in global markets creates pressure for manufacturers to rigidly use automation, and advanced manufacturing techniques to improve manufacturing plant productivity. Among productivity, manufacturers are also interested about other performance dimensions, namely time, flexibility and quality. With the wise use of automation, we could directly improve all of the other indicators, and indirectly quality. In this paper it is shown, how implementation and use of Automatic Guided Vehicles (AGVs) could benefit from the use of manufacturing simulation. AGVs have been available for practical use since the 1970s, but still in the investment and use processes it is rarely seen justified to a high detail, what eventually determines the number of vehicles used in the system. Our paper is able to show with two simulation cases (a hypothetical example and a real case study) that the number of AGVs is very sensitive with respect to production system output. Furthermore, some assumptions that are often made can overly simplify the problem (e.g. no recharging of batteries and constant speed). Therefore, in this paper, we propose that more research should be addressed to the AGV system simulation, and in practice tools for decisions makers should be developed further. Our research also shows that empty travelling and low utilisation is a norm in current AGV systems in order to assure production system functionality, and there is a need and potential for performance improvement here, as well.