Since the seminal work of Schwert (1989a), economic sources of financial volatility have been investigated extensively (e.g., Asgharian et al. (2013); Choudhry et al. (2016); Christiansen et al. (2012); Diebold and Yilmaz (2008); Engle and Rangel (2008); Engle et al. (2013); Nonejad (2017); Paye (2012)). This interest stems from the fact that financial volatility is a crucial input in risk management, portfolio allocation and asset pricing. Financial volatility is also found to successfully predict business cycles, providing early signals of upcoming recessions (Chauvet et al., 2015). However, a recent paper by Paye (2012) shows that although some variables such as treasury spread and default returns can theoretically affect stock volatility, it is difficult to find an individual variable that can predict stock volatility. In detail, adding any macro variables to the benchmark autoregressive model cannot significantly improve out-of-sample forecasting performance. The failure of individual fundamental variables in forecasting stock volatility is further confirmed by more comprehensive analyses conducted by Christiansen et al. (2012), unless some modeling issues such as parameter instability and model uncertainty are addressed (Nonejad, 2017). In this paper, we show that a new variable, crude oil volatility, can be strongly predictive of stock volatility. We demonstrate that oil volatility improves the short-horizon predictability of stock volatility over the autoregressive benchmark model. This predictability is significant during various sample periods. Our investigation complements studies of modeling and forecasting volatility by providing a new fundamental determinant of stock volatility. Our findings are helpful for understanding the economic sources of changes in stock volatility. Various studies have investigated the relationship between oil and stock volatility (see, e.g., Degiannakis et al. (2014); Arouri et al. (2011, 2012); Creti et al. (2013)). Most papers take an in-sample perspective using multivariate GARCH models. However, it is commonly understood that good in-sample performance does not imply that the predictive model displays superior out-of-sample performance. We contribute to the literature by paying attention to the ability of oil to predict stock volatility from an out-of-sample perspective. We employ a parsimonious predictive regression based on realized volatility to provide forecasts, the superiority of which over the GARCH has been well documented in the literature. Our paper is closely related to Driesprong et al. (2008), who find that changes in oil prices predict stock returns in-sample. We complement their work by showing that oil volatility can also predict stock volatility both in-sample and out-of-sample. This paper is also linked to Chen et al. (2010), who show that commodity prices cannot predict asset prices such as exchange rate. We provide a novel example whereby the volatility of a special commodity, crude oil, is shown to have predictive power over stock return volatility. We extend the idea of Chen et al. (2010) to a volatility case and obtain different results.