Reliability optimization problem of complex system is a nonlinear optimization problem under the constraint conditions and in the case of that the objective function has a large number of local extreme values.Hybrid gravitational search algorithm was proposed to solve the model.In the algorithm,a searching mechanism based on the law of gravitation was used to find the global optimal solution.Sequential quadratic programming was employed as a local search method to avoid being trapped into local optimum in the basic gravitational search algorithm.The optimization performance is improved and the search speed is accelerated in the proposed algorithm.Computations on some pratical examples and comparisons with ant colony optimization algorithm,particle swarm optimization algorithm,artificial bee colony algorithm and basic gravitational search algorithm demonstrate the algorithm is feasible and effective.