In a model in which investors have preference forsystematic or individual skewness, we showtheoretically that average skewness shouldpredict future movements in market return. Weextend empirically the work on average volatilityby Goyal and Santa-Clara (2003) and Bali et al.(2005) and use the same data and methodology tostudy realized skewness (i.e., the physicalmeasure of skewness). We find a significantnegative relation between the average stockskewness and future market return. This relationholds for equal-weighted and value-weightedskewness. It holds for our extended sample(1963–2016), which includes the 2007–2009financial crisis, as well as for subsamples. It alsoholds after controlling for the usual economicand financial variables known to predict marketreturns and after excluding firms with smallprice, small size, and low liquidity. Even when ameasure of market illiquidity is introduced intothe regression, the effect of average stockskewness remains significant. In our baselineregression with average skewness alone, a onestandard deviation increase in average monthlyskewness results, on average, in a 0.52% decreasein the subsequent monthly market return.