This study provides evidence on the common determinants for two prominent features of equity market volatility: its persistence over time and its asymmetric dependence on past returns. We show that daily volatility persistence increases with current returns, especially negative returns. It decreases with current volatility. The estimated volatility persistence from the observed variables is termed “conditional volatility persistence”. It provides a new economic link from return to future volatility, and a more robust explanation for their asymmetric relationship. By estimating the variations in the latent volatility persistence, our model significantly improves volatility forecasts relative to recent advances in volatility models.