Offline  handwritten signature  identification is the technique which is used to identification of the behavioural characteristics which is part of biometric characteristic of the each person. But this type of signature identification is subjected to the mal practices, so to prevent this the offline handwritten signature identification using adaptive window position technique, this technique uses the windows for segmentation of the signature image, the windows are of size n*n, the windows are placed on the signature image, after acquiring the position on signature image the window will fragment the images then generates the small sub images, before going to feature extraction phase pattern adjustment technique will applied, this leads to the simple and accurate calculations of the features, there are 10* 50 = 150 signature images are present in the dataset, which are signed by the 10 different persons, each person have signed his signature 15 times, 10 * 10 = 100 signature images are kept in training data set and 10 * 5= 50 signature images are kept in the testing dataset, the comparison between the two sub images takes place based on their similarity properties using an similarity correlation equation, using the adaptive window position technique we can easily find the signature of the genuine user. The accuracy of the proposed method is 94% .