We analyze the impact of financial stress on the predictability of the realized volatility (RV) of five stock markets. To this end, we develop a new volatility model by incorporating the financial stress index (FSI) into the prevailing heterogeneous autoregressive (HAR) model. The empirical analysis demonstrates that the new model significantly outperforms the benchmark HAR model, especially for longterm forecast horizons, suggesting the significant impact of financial stress on the future volatility. The future RV of equity indexes increases with the growth of the FSI. Empirical results hold true for different choices of RV estimators and loss functions.