The mechanisms underlying geocentric (orientations of an object or the body relative to 'gravity') and egocentric estimates (object orientation relative to the 'body') have each been examined; however, little is known regarding the association between these estimates, especially when the body is nearly upright. To address this, we conducted two psychophysical experiments. In Experiment 1, participants estimated the direction of a visual line (subjective visual vertical; SVV) and their own body relative to gravity (subjective body tilt; SBT) and the direction of a visual line relative to the body longitudinal axis (subjective visual body axis; SVBA) during a small-range whole-body roll tilt. We evaluated the correlations between performance on each of these tasks as covariates of actual body tilt angles. Our results showed a significant correlation of performance (estimation errors) on the SVBA task with performance on the SBT task but not performance on the SVV task at the group level after adjusting for the actual body tilt angles, suggesting a link between the estimates for SVBA and SBT tasks. To confirm this relationship, in Experiment 2, we further assessed whether manipulating the subjective direction of the body axis by providing visual feedback in the SVBA task subsequently affected SBT performance. We found that feedback in the SVBA task significantly shifted the SBT angles even when the actual body angles were identical. The observed association between SVBA and SBT performance supports at least a partially shared mechanism underlying body tilt and egocentric estimates when the body is nearly upright.