ABSTRACTIn this paper, we decompose the variance risk premium (VRP) into overnight and intraday components using model‐free implied variance stock indices in the United States, Europe, and Asia. We find that during the nontrading overnight period, the VRP is significantly negative, whereas during the intraday trading period, the VRP becomes positive and often insignificant. We also assess the predictive ability of the overnight and intraday VRPs with respect to future equity returns. We find that the intraday component performs better at shorter prediction horizons, whereas the overnight VRP performs better at longer horizons. Our empirical results suggest that nontrading effects are an important determinant of the VRP.