This study focuses on predictive model for Ra and optimization of cutting conditions in high speed hard turning of X40CrMoV5-1 steel by CBN insert. Cutting speed, feed rate, depth of cut and coating condition have been considered as cutting parameters. The experiments have been performed in CNC lathe according to Taguchi L-32 orthogonal array. The first order mathematical model for the Ra has been developed with multiple regression analysis. Optimization study with analysis of signal-to-noise (S/N) ratios has indicated that feed rate is the negatively most significant factor for Ra under dry cutting conditions. The best surface roughness has been achieved by the lower feed rate and uncoated CBN insert.