The aim of this paper is test for directional predictability between returns and volume (and vice versa) in the energy futures markets. The study employs a cross-quantilogram approach that enables us to assess the temporal association between two stationary time series at different parts of their joint distribution. The data are daily prices and trading volumes from five energy futures markets: West Texas Intermediate (WTI), Brent, natural gas, heating oil and gasoline. The empirical results indicate that high levels of volume lead to extreme returns. Low levels of trading activity have in general no information about future returns. Extreme returns precede extremely high trading volumes for the case of WTI only. The findings shed some additional light regarding traders’ behavior and strategies.