Determining the spread (or the premium) of CDS (Credit Default Swap)s properly is the first step for making the domestic credit derivatives market work. However, as far as we know, there is no literature dealing with won-denominated CDS. We suggest two methods for determining a won-denominated CDS spread and then we analyse the spreads obtained by each method. One method uses the market quotes of dollar-denominated CDS, whose reference entities are dollar-denominated bonds issued by domestic firms. The other method uses the credit spreads of won-denominated risky bonds. In addition, unobservable asset correlation is a critical factor in determining the spreads of basket default swap or a single CDS with counterparty default risk. Hence, we also suggest calculation methods for both implied correlation and historical correlation, and analyse the derived correlation results. Our results are remarkable. First, we can't find any evidence of a long-term equilibrium relationship and price-discovery phenomenon between won-denominated CDS spreads obtained by the two different methods. Second, won-denominated CDS spreads calculated by the first method are shown to be subject to the implied volatility of KOSPI200 options and KOSPI200 returns. However, the spreads calculated by the second method are shown to be subject to the treasury bond yields and the spreads between long-term bond yields and short-term bond yields. Third, the correlation implied by the dollar-denominated basket CDS is quite different from the historical correlation. In fact, the implied correlation is much higher than the historical correlation for basket CDS whose protection seller is a domestic financial institution and protection buyer is a foreign investment bank. Fourth, the historical correlation calculated by using CDS spread data is higher than that calculated by stock return data. The spreads of basket CDSs or a single CDS with counterparty default risk has a negative relationship with asset correlation. Therefore, the CDS spreads using stock returns correlation are higher than those using default swap spread correlation.