The assets under management (AUM) of the Korean National Pension Fund (NPF), which started from 530 billion won in 1988, reached 426.9 trillion won by the end of 2013. This is the fourth largest pension fund in the world. Over 99% of the Korean NPF portfolio is comprised of finance sector holdings and the NPF's performance thus depends on the management of financial sector investments. For many years, major investment vehicles in the NPF portfolio have been domestic bonds, domestic stocks, foreign bonds, and foreign stocks. Recently, alternative investments have become important parts of pension portfolio management. It is common to find alternative investment products such as real estate, infrastructure, private equity, commodities, and hedge funds in various sovereign pension portfolios. As alternative asset classes have little correlation with traditional investments, they help to diversify portfolio risk and also extend portfolios' efficient frontier. In consequence, the proportion of alternative investments among global pension funds increased from approximately 7% in 2003 to 17% in 2012. The Korean NPF has followed this trend. The amount and proportion of alternative investments in the NPF portfolio steadily increased to reach 40 trillion won and to account for 9.4% of the total portfolio as of the end of 2013. However, few studies have been conducted on the management and performance of alternative investments in Korea. This study explores a way to construct the optimal alternative investment portfolio for the Korean NPF using both the Markowitz mean-variance and Black-Litterman models. With six asset classes, we use a proxy for the fund's alternative investment portfolio to test which of the two optimization models is more appropriate for improving portfolio performance. We also construct an alternative investment portfolio with eight asset classes, adding commodity and hedge funds, which are currently excluded from the Korean NPF portfolio, to examine whether including these can enhance the portfolio's efficiency. The main results of this study are as follows. First, we construct the optimal alternative investment portfolio with the Markowitz mean-variance model using the historical average returns of various alternative investment assets as proxy for the equilibrium expected returns. However, the model is extremely sensitive to changes in the input variables and often converges to the corner solution, which allocates unreasonably high weights to one or two assets. This reduces the advantages of portfolio diversification. Second, the Black-Litterman model has been devised to improve these limitations of the mean-variance model. It combines the equilibrium expected returns embedded in a market portfolio with managers' views on its future asset performance. The model alleviates the extreme asset allocation problem. Thus, we suggest that the Black-Litterman model can be more appropriate so long as managers have marketforecasting capabilities. Finally, we confirm that including commodity and hedge funds in the Korean NPF alternative investment portfolio can improve its efficiency, as long as the fund has appropriate market-forecasting capabilities and imposes reasonable portfolio weights restrictions in asset classes. We consider the Black-Litterman model, which reflects managers' views on future performance, in preference to the mean-variance model. We provide new insights into asset allocation methodology for the Korean NPF, which has mainly conducted asset allocation using a Markowitz-type mean-variance model. Some of the current alternative investment segment benchmarks used in the Korean NPF do not explain the variance and covariance of their asset classes, and we propose new benchmarks to replace them. Considering the growing prominence of the alternative investment portfolio in the Korean NPF, we suggest that continuous research effort should be made in this area to further improve the performance of the fund.