Abstract: An method to predict the mechanical properties of welded joints based on adaptive fuzzy neural networks (ANFIS) was studied.By TC4 titanium welded by tungsten inert-gas welding, the tensile strength, bend strength and elongation of welded joints were tested as the data for networks simulation.With the hybrid algorithm of back propagation algorithm and least square algorithm, the model of adaptive fuzzy neural networks to predict mechanical properties of welded joints was established.The results of network simulation show that, the average error is far from less than 7%;according to welding parameter, the mechanical properties including tensile strength, bend strength and elongation can be predicted more accurately by this model;it has the merits of building model easily, simple structure, high precision and good generalization.Consequently, this method can provide an effective approach to estimate mechanical properties of welded joints.