X-ray imaging is an extensively used intra-operative imaging modality because of its low cost and portability. These images have to be corrected for geometric distortion in order to use them for quantitative analysis. Current distortion correction techniques, which use a grid pattern phantom, are not user friendly, for example the phantom interferes with the patient anatomy being imaged. We present a novel method to estimate the c-arm distortion parameters along with the patient pose using patient CT as a fiducial. This method could be very useful in applications such as 3D reconstruction and surgical navigation. In this method, we characterize the c-arm distortion patterns statistically using principal component analysis. The distortion correction method optimizes the distortion parameters by comparing the fluoroscopic images and the projections of the registered patient CT. Our simulation experiments show that the distortion parameters can be recovered up to an average accuracy of 0.5117 mm with a pose error of about 0.1692 mm in translation, 0.2112 deg in rotation.