The critical success of online games has led the industry to global success, as the market size is expected to reach 18,194 USD by the year 2020. However, the success of the online game market has led to the growth of illegal activities, such as the use of game bots. Game bots are software applications capable of collecting game items, which are often banned from online game service providers. The illegal activities are not limited to tax evasion and money laundering. In order to help detect these bots, this study employs the dataset from an MMORPG called Aion. By detecting the bots using the server-side analysis, this paper analyzed user behavior and used features based on the experience, skill, and gravy value that represents the cost-efficiency. We experimented with machine learning algorithms such as MLP, SVM, and Random Forest. As a result, the F-score for detecting the total sum of the accounts that consists of the game bots and real users reached 0.9638. We believe our study may help online game service providers, future researchers, and governmental agencies to detect and classify the MMORPG game bots.