Objective As a reliable alternative to traditional thoracic surgery, thoracic endovascular aortic repair ( TEVAR) has been widely used. During the operation, however, the vascular is invisible. Once the stent graft covers the important aortic branch arteries, serious complications will occur. Therefore, in order to guide the graft, we proposed a preoperative computed tomographic angiography ( CTA) and intraoperative X-ray image registration algorithm for TEVAR. Methods Firstly, using high-density sampling parameters, we generated two DRR sequences of CTA and of skeleton segmented from CTA. Overlapping the corresponding CTA DRR andskeleton DRR, we generated another DRR sequence called DRR library. Then, the parameter which maximized the generalized mutual information between X-ray image and DRR in the library was chosen with Hill Climbing as the registration result. Results Using overlapped DRR librarybased on this method, a CTA image and an X-ray image acquired from a patient in TEVAR were aligned.Compared with generalized cross-correlation based method and DRR library based on this method without skeleton, our approach got more accurate result. Conclusions Overlapped DRR library based on algorithm is valid for CTA and X-ray image registration in TEVAR.