Although hydropower and thermal power constitute the two pillars of China's power supply, the relationship between them has always been a controversial topic. Based on the panel data of China's provincial level covering the period 1991-2015, this paper tried to reveal the relationship between hydropower and thermal power under the framework of autoregressive distributive lag (ARDL) model. Taking the panel unit root and panel cointegration tests as the premise, the PMG estimators implemented under the maximum likelihood method provide unbiased results for this paper. Collectively, In the long run, China's hydropower and thermal power complement each other in ensuring a stable power supply, whereas in the short run, they are competitive to a certain extent, even if factors such as precipitation, grid infrastructure and economic conditions are taken into account. The Granger causality test further reveals the one-way relationship between hydropower and thermal power in short run. That is, thermal power is dominant while hydropower is subordinate and passive, giving an empirical interpretation for the specific phenomenon that discarded water from hydropower plants in China during seasonal water-rich period which has been seriously concerned by policymakers and researchers and appeal for impartial dispatching mechanism. (C) 2019 Elsevier Ltd. All rights reserved.