The design process of journal bearings of turbomachines is complex and time-consuming due to the many geometric and physical variables involved. This paper reports on the design of experiments (DOE) and the response surface design of experiments (RSDOE) methods employed on the design of the drive-end and free-end three-lobe journal bearings supporting a centrifugal compressor rotor. The suitability of each technique is discussed. The bearing design variables employed are bearing slenderness ratio, radial clearance, preload, and lubricant inlet temperature. The rotordynamic response variables selected were the critical speed location, the vibrations at critical speed and operating speed for both bearings, and the threshold speed of instability. The use of a nonlinear (quadratic) RSDOE model is justified. An optimization approach combining an SRDOE and rotordynamic finite element modeling is presented. This method leads to arrive to a multivariate model for multi-objective optimization with very few computations. Identification of the dominant design variables and their effects on several response variables allows establishing engineering feasible solutions with focus on manufacturing versus operating conditions tradeoff.