<p>Meandering is a common phenomenon associated with alluvial rivers. A meander forms when the flowing water in a stream changes its speed, causing erosion of sediment from the outer edge of a bend and subsequent deposition on the inner edge of the bend. A river comprises a meandering main channel and an adjacent flood plain forming a meandering compound channel. Numerical models developed to estimate discharge typically involve solving complicated nonlinear equations. The traditional method gives unpleasant results with significant errors. Various Artificial intelligence approaches are gaining popularity for solving these complicated non-linear problems. Machine learning techniques like Adaptive neuro-fuzzy inference system (ANFIS), Artificial Neural Network–Particle Swarm Optimization (ANN-PSO), and M5 Tree have been utilized in the present study to develop models and estimate discharge in meandering compound channels. The result demonstrated that all three models (ANN-PSO, ANFIS, M5Tree ) effectively estimate discharge in meandering compound channels. The value of R² for all models (ANN-PSO, ANFIS, M5Tree) is 0.973, 0.979, 0.960 respectively and the Root mean square error (RMSE) value for all models (ANN-PSO, ANFIS, M5Tree) is 0.735, 0.695, 0.801 respectively. Based on the result of various statistical indices, the performance of the ANFIS model is found better than ANN-PSO and M5 Tree models. These findings validate the suitability of the proposed ML approaches for discharge prediction in meandering compound channels. Therefore these models recommended accurate estimation of flow parameters in similar hydraulic engineering applications.</p>