There is a strong similarity between the camera pinhole model and the C-arm imaging model of the angiography system. The difference is that the intrinsic parameters of the latter change when C-arm moves because of its special mechanical structure. A new method is presented for the calculation of the extrinsic and intrinsic parameters of a C-arm angiography system from the orientation parameters provided by its output device. This method is built on the imaging principle of angiography system and dynamic geometry of the C-arm, and the model description is based on the related theory in computer vision. Three models are provided, which give different accuracy, complexity and versatility. The first model can be used to calibrate the parameters in selected orientations of the C-arm, the second one in arbitrary orientations of the C-arm when the source-to-image intensifier distance (SID) is fixed, and the third one in arbitrary orientations of the C-arm with arbitrary values of SID. Experimental results show that the first model provides sub-pixel precision and can be used in high-accuracy applications. The accuracy of the second and third models is slightly lower, but can be used in applications with high requirement of flexibility.