The invention discloses an online handwritten signature verification method based on dynamic feature subareas. The method comprises realization steps: (1) a handwritten signature of a user is pretreated; (2) a benchmark signature is determined; (3) the most stable features are selected; (4) length normalization is carried out on the handwritten signature; (5) the distance mean value and the variance between all handwritten signatures and a template in all subareas are acquired; (6) discriminant boundary values of all handwritten signatures in all subareas are calculated; (7) the most stable subarea is determined; and (8) authenticity of the handwritten signature of a to-be-verified person is recognized. The most stable features are used for carrying out length normalization on the handwritten signature, and the best verification result for each person can be obtained. A signature subarea method is adopted, the most stable subarea is dynamically selected for verification according to a different handwriting habit of each person, and the signature judgment accuracy is improved.