This paper tries to estimate transaction cost in online trade. For this, we assume that the changes in transaction price reflect transaction cost. With this assumption, we derived some equations which calculate transaction costs and applied those equations to two exchange sites for game money. In the sites, game money for Lineage(MMORPG game) are exchanged and transaction data were stored. In case of Lineage, 42 servers are operated and we select 3servers from them. The database used includes all actual transaction prices and exchage amount related to those three servers at every day in the year 2010 and 2011. To estimate the transaction cost, we categorized transaction cost into three types: cost for searching transaction opportunity, cost for negotiation, and monitoring and enforcement cost. In the exchange sites, most people just search and accept the asking price. Hence, we assumed cost for negotiation as zero. In case of monitoring and enforcement cost, commission to the sites and cost for settlement are included. The commission was reported by the sites and the settlement cost was not analyzed in this paper. Hence, the main concern of this paper is cost for searching transaction opportunity. The searching cost can be different according to the actual behavior of players and we categorized those into three types; searching asking price,comparing asking price among sites, and suggesting asking price and waiting cost. To derive methodology for measuring transaction cost, some assumptions are used. The first and most important assumption is that the intrinsic value of the game money does not change in short time, at least within one day. By this assumption, all changes in the actual price can be recognized to reflect the transaction cost. Second, among three types of searching costs, cost for ‘suggesting asking price and waiting cost’ is highest, the maximum value. In the case,the one who suggest the price accepts most transaction cost, and the opposite accept less favorable suggested price instead of avoiding transaction cost. In other cases, the players divides the maximum transaction cost. Hence, we assume that the maximum transaction cost is the total search cost for both players. Third, the commission to the site, which is paid by the seller, is reflected to the price and we assume that it is actually paid by both players jointly. Hence, the total transaction cost for both players includes ‘suggesting asking price and waiting cost,’commission, and cost for settlement. In this analysis, cost for settlement was not analyzed. Using those assumptions, we derived some equations for measuring transaction costs and applied them to the data described above. The result is as follows. The maximum search cost,suggesting asking price and waiting cost, was estimated as 11.33% of the traded amount. If we add the commission (3.8%) to the value, the sum becomes 15.13%. This value does not include ‘settlement cost.’ Furthermore, considering the assumptions, the value is underestimated. Considering the estimated value, the transaction cost is considered to be still very high relative to the traded amount even though it is online trade and the object is standardized information good. This result shows the importance of enhancing efficiency with respect to transaction cost. Specifically, the level of commission is less than 25.12%(3.8%/15.13%) of total transaction cost considering underestimation and factors which are not estimated. Hence,by making more sophisticated trade system, the users and the sites can be better results;higher revenue and less transaction cost.