Two dominant theories explaining firms' different level of cash holdings are the static trade-off model and the pecking order theory. According to the former, a firm holds cash and cash equivalent assets when the subject firms' managers can easily enjoy the incentives from the transactions of such assets and the precautionary motive outweigh the cost and benefit of cash holdings. In contrast to the static trade-off model, the pecking order theory asserts that the level of firm's cash holdings is determined passively and that the optimal level of cash holdings does not exist since a firm first expends on the profitable investment opportunities and then pays off the debt and accumulates cash holdings with what is left (Myers and Majluf (1984)). Most empirical studies have also placed emphasis on the static trade-off model and the pecking order theory. Both models exhibit exceedingly similar results that smaller-sized firms tend to increase the level of cash holdings, implying a greater degree of information asymmetry existing in a subject firm especially when its investment and R&D expenditures are projected to be high. Higher expenditures subsequently lead to higher future capital requirement and higher possibility of liquidity shortage owing to more volatility in past cash flows (Opler et al. (1999), Gong (2006)). On the other hand, as Myers and Rajan (1998) have mentioned, managers in a firm with a higher possibility of an agency problem due to the separation of ownership and management are likely to have an incentive to hold more cash assets. This is because cash and its equivalents are the easiest types of assets with which managers can exercise their discretional power.. Conversely, at a firm with more transparent corporate governance, on the other hand, its management has less incentive to hold more cash assets as a way to expand his/her discretional power since, then, such firm closely monitors management's transactions, which reduces the possibility of the agency problem. As such, based on the previous studies, we examine whether the agency theory, as well as the static trade-off model and pecking order theory, is also a significantly influential factor in deciding the level of cash holdings for Korean companies. Most studies consider only partial elements of corporate governance such as ownership structure including shareholding ratios of management, or institutional investors and foreign investors as proxy variables for the possibility of agency problem. Unlike those preceding studies, we use a unique data set of the firm-level corporate governance scores on an annual basis as a proxy variable for the degree of agency problem. The data are being provided by the Korea Corporate Governance Service (KCGS) every year, following the evaluation of listed firms' ownership structures and governance structures. In addition, we provide an empirical evidence on how the agency problem affects a firm's level of cash holdings based on our analysis of 2,541 sample firms listed in the Korea Stock Market between 2002 and 2006. We collect an individual firm's annual financial statement data, stock price data, and composite data with respect to Korean business groups (aka. chaebols) from KisValue II and FnDataGuide. The annual ratio of average cash holdings to total assets ranges from 9.34% to 10.42%. We also consider the standardized score of shareholder protection, board of directors, corporate disclosure (transparency), auditing organization, and earnings distribution which altogether constitute the total corporate governance score. By controlling various factors of the static trade-off model and the pecking order theory, we find that firms with sound corporate governance, especially those that enforce strong shareholder protection rules, tend to reduce the level of cash holdings. Moreover, we also undertake various robustness tests to control for the effects of endogeneity problems of corporate governance and the characteristic of panel data. Furthermore, we provide additional evidence reinforcing our results and the agency theory. We conclude that chaebols show lower levels of cash holding ratios than non-chaebols do, if all else equal. The tendency of the lower the level of cash holdings the better corporate governance the firm has is found more apparent in chaebol firms than in non-chaebol firms. Further, as the level of cash holdings of the other group-affiliated firms increases, the firm's cash holding ratio becomes lower, implying that the static trade-off model and the pecking order theory have less influence on the cash holdings due to the existence of internal capital markets within Korean chaebol groups. In other words, internal capital markets within chaebols indeed have a substantial impact on the group-affiliated firms and allow them to maintain lower level of cash holdings than those of non-chaebols, all else equal. Overall, this paper provides the evidence strongly supportive of the agency theory and shows that differences in the intensity of agency problem across firms play an important role in explaining the different levels of firms' cash holdings.