During the construction process of steel truss arch bridges， it is difficult to calculate the forces of the buckle cable for consideration of both uncertainties and control criteria. In order to solve these problems， a reliability-based buckle cable force optimization method based on multi-objective optimization method is proposed. This method uses support vector regression model as the agent model to fit the relationship between random variables and each target variable， combining with First Order Second Moment method to solve the reliability. Multi-objective particle swarm algorithm based on adaptive grid method is used to optimize the target variables. Compared with the single objective optimization method，this method considers more comprehensive factors. The Pareto optimal solution can be further screened by decision makers based on experience. A plane truss structure is taken as an example to verify the effectiveness of the method. Several reliability calculation methods are compared in terms of advantages and disadvantages. This method is applied to a steel trussed arch bridge with a span of 70 m+240 m+70 m. Both the alignment and stress is optimized， and the reliability of buckle cable is within an acceptable range by optimizing the buckle cable forces.