Statement of problem : It is difficult to obtain a good titanium casting body using the traditional sprue design because of high melting point of Ti, and the low fluidity and high reactivity of molten Ti. Purpose : A new sprue design for titanium casting bodies needs more trial and error. In order to decrease the number of trial and error, computer simulation(MAGMASOFT, Magmasoft Giessereitechnologie GmbH, Achen, Germany) was used to optimize sprue design in U-shaped implant superstructures. Material and method : Five kinds of sprue were examined for the design of the sprue former for titanium casting: Sprue design A(sprue length 4 mm, rectangular shape, 4 sprues), Sprue design B(sprue length 4 mm. round shape. radius 2 mm, 7 sprues), Sprue design C (sprue length 2 mm, round shape, radius 2 mm, 7 sprues). Sprue design D (sprue length 2 mm, cone shape, large radius 3mm. small radius 2mm, 7 sprues), and Sprue design E( sprue length 2 mm. one unit channel shape). Sprue design F(sprue length 2mm, one unit channel shape) was also examined for the design of the customized sprue former in the Biotan system(Schutz Dental Gmbh, Germany). The casting bodies were taken in Sprue design A, Sprue design D, Sprue design E, and Sprue design F in the Biotan casting system. The numerically predicted defects were compared with the experimental dental castings by the radiographic and sectional view observations. Results : 1. According to the result of computer simulation, turbulence during mold filling was decreased in the sequence of Sprue design F, Sprue design E, Sprue design D, Sprue design C, Sprue design B, and Sprue design A. 2. The calculated solidification time contours indicate that hot spot was moved from the casting body to the sprue button in the sequence of Sprue design A, Sprue design B, Sprue design C, Sprue design D, and Sprue design E. The filling pattern of Sprue design F was similar to that of Sprue design E. 3 The predicted filling pattern shows that less turbulence was found in the customized sprue former than in the standard sprue former. 4. According to the results of the radiographic and cross sectional observations, casting defects less than 1mm were found at the center of a casting body with Sprue design E and Sprue design F. However, larger casting defects of 4mm were found in a casting with Sprue design A. 5. The predicted casting porosity was similar to that of the real casting. Conclusion : One unit channel-type and customized sprue former can be recommended. Further research and developement of various sprue designs using computer simulation in necessary to optimize casting design, in order to reduce the formation of casting defects in implant titanuim super-structures.