Although initial public offerings (IPO) are one of the most important issues in corporate finance, academic research in this area has been rather minimal. The absence of academic interests can be accounted by the prevalent assumption in the market that IPO is just a standard procedure toward a firm's growth. The view was not challenged until the early 1990s when Pagano (1993) and Zingales (1995) presented their seminal works. Then a recognition has been built that entrepreneurs actually make a conscious decision whether to go public by evaluating the benefits and costs of going public. Further, it is difficult to collect detailed information such as accounting and ownership data for private firms. For this paper, we analyze the determinants of IPOs using a large database of the private firms that meet the listing requirements of the Korea Exchange (KRX). In doing so, we basically follow the approach of Pagano et al. (1998) by comparing the ex ante and ex post characteristics of IPO firms with those of private firms. First, our main variable in the analysis is the ownership structure which differentiates the paper from the existing ones. As the prior and post IPO ownership structure would be one of the main concerns of the controlling shareholders, we conjecture that owners when designing the sale of new shares will definitely have the final ownership structure in mind. Second, this paper focuses on firms that are controlled by their founders, or the founders' families since they are the ones who are most likely to face the tradeoff between the benefits and costs of going public. Accordingly, family firms account for 95.3% of our sample. In this study, we hypothesize that (1) the probability of an IPO increases with the ownership by the controlling family, but decreases with the ownership by member firms in a business group, and (2) this relationship is stronger for non-chaebol firms than for chaebol-affiliated firms. These hypotheses are built upon the common understanding that less diversified owners are more likely to resort to the IPO, since they have more to gain from diversifying their investments. We test the hypotheses by studying 491 non-financial firms that went public through the KOSPI market of the KRX during the period of 1987 through 2005. To investigate the determinants of Korean IPOs, the characteristics of these IPO firms are compared with those of non-IPO firms that still met the listing requirements of the KRX but chose to remain private. In addition, we investigate what types of changes resulted from going public especially in their financial characteristics and the ownership structures . We also analyze the determinants of changes in the ownership by the controlling family over the three years before and after the IPO. Our findings are summarized as follows. ∙ Ex ante determinants of an IPO: From a logit model of the probability of going public, we find that firms are more likely to go public in the following cases: if the median market- to-book ratio of listed firms in the same industry is higher; and if the ownership by the controlling family is higher while the ownership by member firms in the same group is lower. We also find that these results are stronger for non-chaebol firms than for chaebol firms. The overall results support our hypotheses that controlling shareholders decide to go public in order to diversify their own wealth and maximize the proceeds by carefully timing the IPO. ∙ Ex post consequences of an IPO: By examining the dynamics of firm characteristics before and after the IPO through a regression framework of Chemmanur et al.(2010), we find that the following firm characteristics notably decline after the IPO: profitability, sales growth, investment, leverage, the industry market-to-book ratio, family ownership, and the ownership by affiliated firms. The ex post effects of the IPO complement our results for the ex ante analysis. Subsequent to the IPO, investments in tangible assets also decrease while investments in affiliated firms' equity increase, especially for non-chaebol firms. This result implies that the proceeds from the IPO are not used for future investments, but to diversify the business group or secure control over the group of a controlling shareholder. ∙ Determinants of the changes in ownership by the controlling family: Finally, we run each regression for the two separate periods before and after the IPO to examine the determinants of changes in the ownership by controlling families. For the periods prior to the IPO, the ownership by a controlling family is positively related to the investments in affiliated firms' equity. After the IPO, this relationship is no longer significant, and family ownership increases with the Tobin's Q. These results suggest that controlling shareholders maximize their private benefits by securing their control over business groups before the IPO, but pursue higher investment value after an IPO when the control issue tends to be less important.