When entering a coordinated flight turn without visual references, the perception of roll-angular displacement is determined by vestibular cues, and/or probably by assessment of the G load (G magnitude) and its translation into the corresponding bank angle. Herein, we examined whether repeated exposures to hypergravity (G training) in a centrifuge, would advance, not only the ability to accurately assess the G load, but also the capacity to detect or estimate the corresponding roll inclination of the centrifuge gondola. To this end, in 9 men without piloting experience, the subjective estimation of G load and roll tilt were assessed, in complete darkness, during 5-min coordinated turns in the centrifuge, performed at 1.1G (25° roll-tilt angle) and 2.0G (60° roll tilt angle). These trials were conducted before and after 5-weeks of G training [3×40-min sessions・week-1; protocol: 20×1-min at G levels close to the individual relaxed G-level tolerance (range: ∼2.6G(~67°)-3.6G(74°)), separated by 1-min intervals at idle speed (1.4G)], while continual feedback to the subjects was limited to the G load. As expected, G training improved subjects' capacity to assess G load, especially at 2.0 G (P=0.006). The perception of roll tilt, however, was consistently underestimated (by ~70-80%), and not enhanced by G training (P≥0.51). The present findings demonstrate that prolonged repeated G-induced roll-tilts in a centrifuge gondola, while external feedback is restricted to graviception, enhance the capacity to perceive G load, but fail to advance the ability to detect or consciously estimate the magnitude of roll-angular displacement during a coordinated turn.