Natural guidewire manipulations of surgeons in percutaneous coronary intervention (PCI), rotating and translating a guidewire with finger and hand motion, are important and useful references to build a knowledge base (KB) for robotic autonomous intervention. In this paper, guidewire manipulation models are proposed using an improved data glove, in which a 6-DOF position/orientation sensor is installed. The tracking functions between the surgeon’s manipulations and guidewire motions are established according to the models. The performance of proposed method is analyzed with tracking errors and the statistical significance of tracking trajectories is evaluated with the Kruskal-Wallis test. The results show that radial rotation tracking error is 20.24 ± 8.38 (degree) and axial advancement tracking error is 0.99 ± 0.81 (mm) at a medium delivery speed.