We present a new range image segmentation algorithm based on the extraction of planar and general quadric surfaces, which is of great relevance in man-made environments, built largely of low-order surfaces. We describe how our robust estimator can effectively reject erroneous surface hypotheses, while identifying points describing a general quadric surface, by means of an improved approximation to the true geometric distance between each point and the surface. We present thorough experimental results with quantitative evaluation against ground truth.