The rapid growth of hedge funds in recent years has been accompanied by cases of severe failure. Since the aftermath of Long Term Capital Management (LTCM), investors recognize that hedge funds may provide high expected returns but they might be exposed to a huge downside risk that is not easily detected by traditional risk measures. My dissertation consists of three essays on the risk of hedge funds. In the first essay, I use a cross-sectional approach to analyze the risk-return trade-off. I compare semi-deviation, value-at-risk (VaR), Expected Shortfall (ES) and Tail Risk (TR) with standard deviation. Using the Fama and French (1992) methodology and TASS data, I find that the left-tail risk captured by Expected Shortfall (ES) explains the cross-sectional variation in hedge fund returns very well, while the other risk measures provide insignificant results. During 1995-2004, hedge funds with high ES outperform those with low ES by an annual return difference of 7%. In the second essay, I implement a survival analysis based on the Cox proportional hazard (PH) model to compare downside risk measures with standard deviation in predicting hedge fund failure. I find that funds with high ES have a high hazard rate when controlling for the style effect, performance, fund age, size, lockup, high-water mark (HWM) provision, and leverage. Standard deviation, however, loses the explanatory power when the other explanatory variables are included. As I find that liquidation does not necessarily mean failure in the hedge fund universe, I suggest simple criteria such as the last six-month return and change in fund size rather than the stated drop reasons to calibrate hedge fund failure. I reexamine the attrition rate of hedge funds based on these findings and argue that the real failure rate of hedge funds (3.1%) is lower than the attrition rate (8.7%). In the third essay, I focus on the liquidity risk premium. I examine the relationship between hedge fund share restrictions and liquidity premium by comparing offshore and onshore hedge funds. Due to tax provisions, offshore and onshore hedge funds have different legal structures, which lead to differences in share restrictions. On average, offshore funds impose less share restrictions, hence they underperform their onshore counterparts due to illiquidity premium, consistent with Aragon (2007). However, I find that once offshore funds impose share restrictions the illiquidity premium is higher because of a tighter relation between share illiquidity and asset illiquidity in offshore funds. Introducing the lockup provision increases the abnormal return by 4.4% per year for offshore funds compared with only 2.7% for onshore funds. I also find that share illiquidity premium becomes lower when an offshore fund is affected by its onshore equivalence through a master-feeder structure.