Segmentation of meaningful objects from intensity volume data and shaded surface display of the extracted objects are fundamental operations in the visual evaluation of volume images obtained in medicine by, e.g., computed tomography (CT) and magnetic resonance (MR) imaging. A fast algorithm to generate shaded surface views directly from voxel data is presented. Even noncubic voxels with the shape of a general parallelepiped as typically measured in CT are accepted as input without preinterpolation. The display algorithm has been modularized and implemented in an inexpensive microprocessor system currently equipped with four processors which operate in parallel. The mere software implementation renders an oblique view from any surface of a data set of 2563 voxels in a few seconds including on-line interpolation to cubic voxels, thresholding, gradient shading, and effects of clipping planes. The original voxel intensities are accessible at any time during the display process; they can also be combined with the surface display for an improved visualization effect. An efficient and robust segmentation method for the extraction of objects from complicated medical volume data is described. The method consists of 3D edge and surface detection, object refinement by matched filters or morphological operations, and a 3D connected component labeling procedure enabling interaction to change the classification of the automatic algorithm. Results from applying the segmentation and display techniques to CT and MR data are presented.