•We analyze the role of financial stress in forecasting oil-price volatility.•We use various variants of the Heterogenous Autoregressive model of realized volatility.•We find that indexes of financial stress help to improve forecasting performance.•The shape of the forecaster loss function that used to evaluate performance is important.•Alternative types of investors benefit from monitoring different regional sources of financial stress. We analyze the role of global and regional measures of financial stress in forecasting realized volatility of the oil market based on 5-min intraday data covering the period of 4th January, 2000 until 26th May, 2017. In this regard, we use various variants of the Heterogeneous Autoregressive (HAR) model of realized volatility (HAR-RV). Our main finding is that indexes of financial stress help to improve forecasting performance, with it being important to differentiate between regional sources of financial stress (United States, other advanced economies, emerging markets). Another key finding is that the shape of the forecaster loss function that one uses to evaluate forecasting performance plays an important role. More specifically, forecasters who attach a higher cost to an overprediction of realized volatility as compared to an underprediction of the same absolute size should pay particular attention to financial stress originating in the U.S. But, in case an underprediction is more costly than a comparable overprediction, then forecasters should closely monitor financial stress caused by developments in emerging-market economies. In sum, financial stress does have predictive value for realized oil-price volatility, with alternative types of investors benefiting from monitoring different regional sources of financial stress.