This project is a continuous research of a topic well-known in the literature, namely, the performance study of unbalanced unpaced production system. In the literature, there were many studies that investigated the statistical outputs of an unbalanced production line using simulation. This project focuses on researching the outputs like average buffer level and idle time that are rarely studied in previous research by using optimization tools from discrete event simulation software FACTS.The models used in the article (Shaaban & McNamara, 2009) have been used as a guideline during the development of the simulation models for this project. Two simulation models were created, each using different discrete event simulation software, namely FACTS analyzer and Plant simulation. Those simulation models fulfills its role in verification & validation stage, with their statistical outputs compared to each other and with Shaaban and McNamara’s results. After verification & validation comes optimization of those simulation models, by using optimization tools from FACTS.The research area expanded during the optimization phase. Originally Shaaban et.al analyzed unbalanced production line with one fixed value of coefficient of variation. In order to expand the view on the properties of an unbalanced production line, three more coefficient variation were added with total of four in this project. As a result, 12 optimization results were created at the end of this project. Each optimization has 30 000 iterations to ensure its convergence.The first step of analysis is done by locating all Pareto-optimal solutions with optimization tools in FACTS. The raw data of all solutions are later transferred and converted into EXCEL files. Using scatter graph and putting all outputs against each other in EXCEL, it creates visual graph that can be used to analyze and to investigate interesting behavior in an unbalanced production line.The analysis on the optimization results showed several interesting behaviors from production line with different settings. One being that if a production line possess worse coefficient of variation than its competition. By raising the inter-stage buffer level of the production line with inferior coefficient of variation, it can achieve the same level, if not greater outputs than its competitor who possess better coefficient of variation. The other interesting behavior are optimization results with highest outputs in regard of either idle time or average buffer level, with deep analyzation using optimization tools from FACTS. Certain operation time pattern and inter-stage buffer pattern could be observed from those results.