A novel approach of principal lines ex traction on the whole palm is proposed to solve the problems of the light int erference,the complexity of algorithm and the inaccurate region of interest (RoI) positioning in traditional extraction algorithms of princi pal lines. This algorithm is based on gray level image and uses the prior knowledge of the direction certainty,gray-scale similarity and start po ints′ and end points′ regularity of principal lines to extract the five classes of palm-print principal lines (emotion line,intellige nce line,life line,career line and success line).It effectively avoids the complex computing on images enhancing,ROI positioning and palm-prin t image binaryzation in traditional methods,and retains the information about principal line′s direction,relative position,widt h and the number of principal lines.The method gets principal lines extraction r ate(ER) of 95.6% by the test of HKUST,and the influence of translation,rotation and light is small. Compared with the existing algorithms,this method has obvious advantages of lower complex ity,higher ER,retention of the information quantity and integrated information,and provides supports for the re search of multi-model hand recognition.