This paper provides evidence that supports the original hypothesis of Chordia, Subrahmanyam, and Ashuman (2001) that greater variability in liquidity should lead to higher expected returns. While prior research has often found a negative relation between the volatility of liquidity and expected stock returns, we find that the volatility of the bid-ask spread is positively related to future returns. The average risk-adjusted return for stocks in the highest spread volatility quintile is around 1.7 percent per month, with returns from High-Low quintiles as high as 2.7 percent per month. Furthermore, the spread volatility premium is robust to a variety of multivariate tests that control for the market risk factor, SMB, HML, momentum, and illiquidity risk. Our findings provide support for the hypothesis that variability in liquidity affects expected returns and is an important component of illiquidity.