In this paper, based on Acharya and Pedersen’s [Journal of Financial Eco- nomics (2006)] overlapping generation model, we show that liquidity risk could influence the market risk forecasting through at least two ways. Then we argue that traditional liquidity adjusted VaR measure, the simply adding of the two risk measure, would underestimate the risk. Hence another approach, by modeling the liquidity adjusted returns (LAr) directly, was employed to incorporate liquidity risk in VaR measure in this study. Under such an approach, China’s stock market is specifically studied. We estimate the one-day-ahead “standard” VaR and liquidity adjusted VaR by forming a skewed Student’s t AR-GJR model to capture the asymmetric effect, non-normality and excess skewness of return, illiquidity and LAr. The empirical results support our theoretical arguments very well. We find that for the most illiquidity portfolio, liquidity risk represents more than 22% of total risk. We also find that simply adding of the two risk measure would underestimate the risk. The accuracy testing show that our approach is more accurate than the method of simply adding.