This study aims to investigate the threshold effect of GDP on the causality between GDP and energy consumption (EC), using panel data of 26 OECD countries over the period 1971–2014. Threshold regression technique is employed to explore whether there exists a threshold of GDP in the relationship between GDP and energy consumption and Granger causality testing procedure based on panel VECM is applied to test the causality across GDP regimes. The empirical results highlight the existence of GDP threshold in which the effects of GDP on energy consumption and the direction of the energy-growth causality depend on the initial value of GDP. When real GDP per capita is less than US$ 48,170, there exists unidirectional causality from EC to GDP in both the short and long run. However, when GDP exceeds US$ 48,170 per capita, there is no causality relationship in the short run but GDP is found to Granger cause EC in the long run. The findings urge for the attention of policymakers to take the initiative in implementing energy conservation policies when the economy reaches certain high levels of development and to switch energy policies appropriately in different periods of economic growth, to foster sustainable development.