In this paper, we describe the first prospective application of the shape-comparison program, WEGA (weighted Gaussian algorithm), to find new scaffolds for anti-tumor agents. A series of sixteen carbazole alkaloids extracted from Clausena vestita D. D. Tao, which have anti-tumor activities at the cellular level, were used as query molecules. A compound library was screened by ranking molecules based upon their 3D shape and pharmacophore similarities to known inhibitors. The relationship between the structures and activities was also studied through comparative molecular field analysis (CoMFA). Twelve hits show comparable growth inhibition activity against HepG2 cells (a hit rate of 60%); eight of the hits have new scaffolds (in comparison with known inhibitors). These results indicate that a shape-based screening approach, such as WEGA, can be efficiently used for scaffold hopping in a lead identification process.