The paper discusses the first evaluation of a 3D Live-Wire technique that operates on over-segmented 3D mosaic volumes. Pre-processing normalizes data by histogram spreading followed by over-segmentation with region growing. This leads to a robust and intuitive parameter selection for various tissues and image modalities without requiring problem specific knowledge. During interactive segmentation from a set of user-defined seed points a triangle mesh using Delaunay triangulation is generated. For each triangle, defined by three seed points, a closed surface patch is generated by the 3D Live-Wire method. For evaluation purpose white cerebral matter from MRI images is to be segmented. A mean deviation from the correct object boundary extracted from available model data of about 0.60 pixels (+/- 0.23) is observed while the required time was about 229.5 seconds (+/- 173.2).