The surface roughness of workpiece is improved during spark-out. This improvement is due to the overlapping cutting between abrasive and Workpiece. The reduction of the surface roughness during spark-out is promoted by grain sharpness as a result of fracture wear. If the value and the reduction trend of the surface roughness during spark-out can be predicted, the automation and the promotion of efficiency of grinding can be achieved. Therefore the prediction of surface roughness is important. In this study, the prediction of spark-out surface roughness was performed by the computational simulation. The effect of table speed, depth of cut and grain shape on the surface roughness during spark-out could be analyzed by the computer simulation instead of the experiment. The effect of the enlarged overlapping area can be simulated by changing the angle of grain tip. The angle of grain tip can be adjusted by changing dressing lead and sharpness of dressing tool in reality.