River flow regulations and thermal regimes have been altered by human-induced interventions (such as dam construction) or climate change (such as air temperature variations). It is of great significance to adopt a well-performed data-driven model to accurately quantify the impact of human-induced interventions or climate change over river water temperature (WT), which can help understand the underlying evolution mechanism of the river thermal regimes by dam operation or climate change. This research applied the Bayesian network-based model (BNM), which can easily identify inherently stronger associated variables with a target variable from multiple influencing variables to predict the daily WT and make a quantitative assessment of the effect produced by Three Gorges Reservoir (TGR) construction in the Yangtze River, China. A comparative study between the proposed model and two other models was implemented to verify the predicted accuracy of the BNM. With the help of the BNM model, the impact of reservoir impoundment over water temperature was quantitatively analyzed by calculating the difference between reconstructed water temperature series and observed series during the post-TGR period. The construction of the TGR posed more impact on variations in WT than the impact induced by the climate change according to results. The effect of TGR on WT can be concluded as follows: WT from October to January in post-TGR showed a remarkable warming tendency and an increase in released warmer water volumes than before, while WT showed a cooling tendency during March to June because of the hysteretic effect of WT response to increasing air temperature. The proposed BNM model shows great potential for WT prediction and ecological risk management of rivers.