Introduction: Efficient water resource management in irrigation systems relies on the accurate estimation of seepage loss from lined canals. This study utilized machine learning (ML) algorithms to tackle this challenge in seepage loss prediction. Methods: Firstly, seepage flow through irrigation canals was modeled numerically and experimentally using Slide2 and physical models, respectively. Then, the Slide2 model results were compared to the experimental tests. Thus, the model was used to conduct 600 simulation scenarios. A parametric analysis was performed to investigate the effect of canal geometry and liner properties on seepage loss. Based on the conducted scenarios, ML models were developed and evaluated to determine the best predictive model. The ML models included non-ensemble (regression-based, evolutionary, neural network) and ensemble models. Non-ensemble models (adaptive boosting, random forest, gradient boosting). There were four input ratios in these models: bed width to water depth, side slope, liner to soil hydraulic conductivity, and liner thickness to water depth. The output variable was the seepage loss ratio. Seven performance indices and k-fold cross-validation were employed to evaluate reliability and accuracy. Moreover, a sensitivity analysis was conducted to investigate the significance of each input in predicting seepage loss. Results and discussion: The findings revealed that the Artificial Neural Network (ANN) model was the most dependable predictor, achieving the highest determination-coefficient (R2) value of 0.997 and root-mean-square-error (RMSE) of 0.201. The eXtreme Gradient Boosting (XGBoost) followed the ANN model closely, which achieved an R2 of 0.996 and RMSE of 0.246. Sensitivity analysis showed that liner hydraulic conductivity is the most significant parameter, contributing 62% predictive importance, while the side slope has the lowest significance. In conclusion, this study presented efficient and cost-effective models for predicting seepage loss, eliminating the need for resource-intensive experimental or field investigations.