Estimating household water consumption can facilitate infrastructure management and municipal planning. The relatively low explanatory power of household water consumption, although it has been extensively explored based on various techniques and assumptions regarding influencing features, has the potential to be enhanced based on the water-energy nexus concept. This study attempts to explain household water consumption by establishing estimation models, incorporating energy-related features as inputs and providing strong evidence of the need to consider the water-energy nexus to explain water consumption. Traditional statistical (OLS) and machine learning techniques (random forest and XGBoost) are employed using a sample of 1320 households in Beijing, China. The results demonstrate that the inclusion of energy-related features increases the coefficient of determination (R²) by 34.0% on average. XGBoost performs the best among the three techniques. Energy-related features exhibit higher explanatory power and importance than water-related features. These findings provide a feasible modelling basis and can help better understand the household water-energy nexus.