The invention discloses an iterative data detection method based on localized optimization, and belongs to the technical field of computer storage. The problem that the detection efficiency is low in existing iterative data detection methods is solved, and the current situation that as the scale of stored data is expanded, the iterative data detection efficiency is lowered is adapted to. The method includes the steps of Bloom filter detection, Hash bucket write buffer detection, Hash bucket read buffer detection and Hash bucket address table detection. According to the method, mainly focusing on dataset types with high locality, through mining the locality of data concentration, the data prefetching efficiency is improved, the disk access overhead is reduced, and the throughput rate of data deduplication is improved. For probable iterative data of data concentration, the repeatability of a data block is prejudged initially by means of the Bloom filter, then based on different conditions, detection of three-order iterative data is conducted on the hot zone, cold zone and disc of buffer cache respectively, thus the locality of the iterative data is fully utilized, and the detection effectiveness of the iterative data is improved.