This paper presents the rotordynamic performance measurements and model predictions of a rotor supported on three-pad gas foil journal bearings (GFJBs) with various mechanical preloads. The rotor with its length of 240 mm, diameter of 40 mm, and weight of 19.6 N is supported on two GFJBs and one pair of gas foil thrust bearings (GFTBs), being a permanent magnet rotor of a high speed electric motor. Each bearing pad consisting of a top foil and a bump-strip layer is installed on a lobed bearing housing surface over the arc length of 120 deg along the circumference. Test three-pad GFJBs have four different mechanical preloads, i.e., 0 μm, 50 μm, 70 μm, 100 μm with a common radial nominal clearance of 150 μm. A series of speed-up tests are conducted up to 93 krpm to evaluate the effects of increasing mechanical preloads on the rotordynamic performance. Two sets of orthogonally positioned displacement sensors record the rotor horizontal and vertical motions at the thrust collar and the other end. Test results show that the filtered synchronous amplitudes change little, but the onset speed of subsynchronous motions (OSS) increases dramatically for the increasing mechanical preloads. In addition, test bearings with the 100 μm preload show a higher OSS in load-on-pad (LOP) condition than that in load-between-pads (LBP) condition. A comparison with test results for a one-pad GFJB with a single top foil and bump-strip layer reveals that three-pad GFJB has superior rotordynamic performance to the one-pad one. Finally, the test data benchmark against linear rotordynamic predictions to validate a rotor-GFJB model. In general, predicted natural frequencies of the rotor-bearing system and synchronous rotor motions agree well with test data. However, stability analyses underestimate OSSs recorded during the experimental tests.