This paper investigates the eect of the ETF illiquidity on the ETF tracking error, return, and volatility. Both ETF illiquidity and ETF tracking errors are positively related and are persistent over time. The empirical tests of the liquidity adjusted capital asset pricing model show that illiquid ETFs tend to be more sensitive to underlying index returns or ETF market liquidity, and a positive liquidity premium exists in the US ETF markets. Further, I show that an ETF variance is typically larger than its NAV variance when the ETF is not actively traded. This result conrms that illiquid ETFs are much riskier when market liquidity declines sharply.