Humans are able to estimate the vertical direction of an Earth fixed reference frame, which estimate is known as the subjective vertical (SV). To identify the SV, a distinction must be made between accelerations due to self-motion and gravity. Previous studies on this topic measured the SV using a variety of methods possibly affecting the outcome differently. In this study subjects were sinusoidally moved around their naso-occipital axis and their SV was dynamically measured using a joystick. In half the experimental conditions, the joystick was moved with the motion and was kept vertical on other experimental conditions, thus moving against self-motion. Although physically indicating the same angle, the average perceived angle was larger when moving the joystick with the motion than against. The difference can be explained by assuming an idiotropic vector being at issue when measuring the subjective vertical, and not when measuring subjective tilt. © 2012 - IOS Press and the authors. All rights reserved.