This paper is an empirical investigation of the determinants of the cross- section of stock returns in Korea. While Fama and French's (1993) three-factor asset pricing model is known to perform reasonably well in explaining the cross-section of stock returns in the U.S. and many other stock markets in developed countries, the model's performance in explaining the cross-section of stock returns in Korea has been less than satisfactory. More specifically, there is mixed evidence for the existence of the book-to-market effect in the Korean stock market. Because Fama and French's three-factor model is empirically motivated without firm theoretical grounds, it is difficult to argue that the book-to-market factor (HML) is a priced risk factor in the Korean stock market without clear empirical evidence for the book-to-market effect in the cross- section of stock returns in Korea. Consequently, an alternative asset pricing model is necessary to be used widely in Korea for the purpose of risk adjustment in estimating the cost of capital or performance evaluation. Using monthly stock returns and accounting information for the sample of non-financial firms belonging to the KOSPI index over the 1991~ 2007 period, we first investigate the cross-sectional relationship between stock returns and firm characteristics in the Fama-MacBeth regression framework. We find no evidence for the book-to-market or momentum effect. Firm characteristics which show significant relationship with the cross-section of stock returns in Korea are firm size measured by market capitalization and liquidity measured by turnover. Given the evidence for the size and turnover effects, we construct portfolios by sorting firms into four size groups and four turnover groups independently, similar to the way in which Fama and French (1993) construct their size and book-to-market portfolios. The sixteen size-turnover portfolios show clear patterns in average returns, decreasing in both firm size and turnover. Moreover, these size-turnover portfolios exhibit much bigger spread in average returns than the portfolios constructed using firm size and book-to-market (size-BM portfolios), which suggests that the size-turnover portfolios may be more useful and relevant test assets than the size-BM portfolios in the Korean stock market. The clear pattern of average returns in the size-turnover portfolios suggests that liquidity measured by turnover may be a priced risk factor in Korea. In order to estimate the magnitude of the risk premium associated with the size and turnover effects, we construct mimicking portfolios which are designed to capture the effects of firm size and turnover in the cross-section of stock returns, similar to the way in which Fama and French (1993) construct their mimicking portfolios, SMB and HML. Consistent with the results from the Fama-MacBeth regressions, the size factor (SMB) and the turnover factor (NMP) show significantly positive average returns while the average return of the book-to-market factor (HML) is positive but statistically insignificant. The magnitudes of the average returns for SMB and NMP are also economically significant at 0.99% per month for SMB and 1.19% per month for NMP. We then perform standard time-series and cross-sectional tests of asset pricing models using the size- turnover and size-BM portfolios as test assets. The asset pricing models we consider are the Fama-French three-factor model and our alternative three-factor model which replaces the book-to-market factor HML with the liquidity factor NMP. The main findings are as follows. When the size-turnover portfolios are used as test assets, the GRS F-test rejects the Fama- French three-factor model while it does not reject our alternative three-factor model. Moreover, the estimated risk premium for the liquidity factor NMP is both economically and statistically significant at 1.3% per month, which is also similar in magnitude to its average return of 1.19% per month. When the size-BM portfolios are used as test assets, the two models show broadly comparable performance. These main findings remain unchanged when we test the models using the first half or the second half of the sample period. In particular, we find that the liquidity factor NMP shows significant explanatory power for the cross-section of stock returns in each of the sub-sample periods. The Fama-French three-factor model is widely used in research and practice for risk adjustment and performance evaluation in both the U.S. and Korea. But the lack of clear empirical evidence for the book- to-market effect in Korea raises a question about the relevance of the book-to-market factor HML and the Fama-French three-factor model in understanding the determinants of stock returns in Korea. The findings in this paper suggest that our alternative three-factor model which incorporates a liquidity factor may be more useful and relevant for understanding the systematic variation of stock returns in Korea.