The present invention relates to a method identifying watermarks in a PDF document. The method comprises: Step 1: extracting PDF characters, bitmap element content and location information, and calculating and collecting statistics of fingerprints according to the location, wherein the fingerprint is the key information extracted from the content (here is the location) and can represent the pointed content, and the number of times of the same fingerprint can be counted; Step 2: filtering (count & gt; page/2) the fingerprint set, and searching for elements corresponding the fingerprint location; Step 3: traversing the alternative fingerprints, if all fingerprints correspond to the character elements, proceeding to Step 4 for processing, and if all fingerprints correspond to the bitmap elements, proceeding to Step 5; Step 4: extracting the characters corresponding to the fingerprints, and if the characters are the same or meet the regulation of increasing or descending, determining the fingerprints as the text watermark; and Step 5: extracting the bitmaps corresponding to the fingerprints, and if the bitmaps are the same, determining the fingerprints as the picture watermark. In the processing frame, the present invention also discloses implementation details such as the location fingerprint generation method, the character content regulation determination method, the same picture determination method, and the like, so that the identification rate of the watermarks of the PDF document can be comprehensively improved, and preparation is made for the subsequent processing such as content extraction and analysis, and the like.