We have developed an automated image-searching method based on biological fingerprints for identifying correct patients in misfiled chest radiographs in a picture archiving and communication system (PACS) server. We used five biological fingerprints including distinctive anatomic structures in a misfiled chest radiograph of an unknown patient to find another image of the same patient stored with correct patient information in a PACS server. The correlation values were determined for the corresponding biological fingerprints in all images in the image server. The correlation indices as a measure of the overall similarity of the two images were determined from the summation of five correlation values and the combination of correlation values with the weighting factors. Finally, the correct patient was identified automatically by the image with the highest correlation index. By use of the summation of five correlation values as the correlation index, 78.0% (156/200) of the 200 patients for misfiled images were correctly identified in the database. When we applied the weighting factors for each biological fingerprint to determine the correlation index, the performance in identifying the correct patient was improved to 87.5% (175/200). An additional 5.0% (10/200) of images were included in the Top 10 ranking of the correlation index in the database. These cases could be identified manually by radiology personnel. We conclude that the automated image-searching method based on biological fingerprints with weighting factors would be useful for identification of the correct patient in the case of misfiled chest radiographs in a PACS server.