In this paper, we present an approach to estimate camera intrinsic and extrinsic parameters for roundabout traffic scenes in intelligent transportation applications. Unlike many previous works on camera calibration for roadway scenes based on the parallel line method, our work addresses camera calibration at roundabout scenes by using a circle. The proposed method can estimate tilt angle, focal length, and camera height by matching the ellipse equation extracted from an image with the perspective-transformed equation of the corresponding circle. Also, only one image is needed for camera calibration. The method is validated with real world roundabout traffic scenes and the results are comparable to those from the parallel line method in terms of accuracy.