Test node optimization problem is a key problem of diagnostic design for circuit, which belongs to an N-P completeness problem Algorithms being applied for the test node optimization problem have some problems such as slow searching speed and no convergence, so a kind of method based on genetic particle swarm optimization algorithm was presented to solve the problem. It optimizes the test node's “fault-test” matrix by genetic algorithm and searches quickly by particle swarm optimization. Searching with the method is faster than traditional methods, and the optimal results are also better. The method has been used in engineering practice.