Abstract: To analyze both hamming distance and cartesian distance, by improved the test sequences' relevance by the classical cellular automate (CA), optimization of BIST test vectors with pre-determined distance is proposed.The method through achieving maximum of the Hamming distance and pre-setting Cartesian distance to generate pre-determined distance test sequences, then gaining the best pre-determined test sequences by the global optimization operation of genetic algorithm.The results of ISCSA'85 experiment indicates that, the pre-determined test vectors generated by this testing method has higher fault coverage ratio and test efficiency than LFSR-CA structure, which makes the method in testing is more effective in testing hard-detected faults.