In this paper an off-line Chinese writer retrieval system based on text-sensitive writer identification is practised. OCR technique is used to mark the handwritten script content. The used text-sensitive writer identification algorithm extracts directional element features(DEFs), reduces the dimensions using PCA and LDA, and adopts the simple Euclidean classifier. By introducing negative samples, the similarity measure is more comprehensive. However the writer identification algorithm is text-sensitive, a novel method for retrieving in an amount of writers who has different handwriting script content is proposed in this paper, by combining and sorting the confidence results of writing identification. An experiment, which is carried out in a handwriting script database, demonstrates the effectiveness of the proposed method.