Recently, the polishing process using magnetorheological fluids(MR fluids) has been focused as a new ultra-precision polishing technology for micro and optical parts such as aspheric lenses, etc. This method uses MR fluid as a polishing media which contains required micro abrasives. In the MR polishing process, the surface roughness and material removal rate of a workpiece are affected by the process parameters, such as the properties of used nonmagnetic abrasives(particle material, size, aspect ratio and density, etc.), rotating wheel speed, imposed magnetic flux density and feed rate, etc. The objective of this research is to predict MRR according to the polishing conditions based on the multiple regression analysis. Three polishing parameters such as wheel speed, feed rates and current value were optimized. For experimental works, an orthogonal array L27(313) was used based on DOE(Design of Experiments), and ANOVA(Analysis of Variance) was carried out. Finally, it was possible to recognize that the sequence of the factors affecting MRR correspond to feed rate, current and wheel speed, and to determine a combination of optimal polishing conditions.