Abstract: An integrated model of rough set and neural network (RS_RBFNN) was proposed for predicting fatigue life of titanium alloy welded joints. The fatigue data were discretized by using the entropy-based algorithm, and the fatigue evaluation indices were reduced without information loss through a genetic algorithm. The reduced indices were used to develop the rules for fatigue life of welded joints and to train the RBF neural network. The rough set theory was used to determine the category of fatigue life for the test samples which matched the rules in the rule-base. The neural network was applied to those test samples which did not match any rules in the rule-base. Experimental results based on the fatigue data of titanium alloy show that the RS_RBFNN model for fatigue analysis of welded joints had improved fault tolerance and precision. Therefore this model is of practical significance for predicting fatigue life of titanium alloy welded joints.