We tend to following moving objects with our eyes. To estimate their velocities, therefore, we must take account of our eye movements. During smooth pursuit, velocity judgements can be led astray by moving the background. Do we misjudge an object's velocity when the background moves because the additional shift of the background's image on the retina is interpreted as the result of additional motion of the observer rather than as motion of the background? In the present experiment, the traditional configuration of target and background was supplemented with a "floor of tiles" drawn in perspective directly under the "background". The motion of this new simulated plane was used to specify whether the additional retinal shift represents actual motion in the background, rotation of the observer's eyes, or observer locomotion parallel to the target. Moving the background clearly influenced the perceived velocity of the target. However, "specifying" whether the observer or the background had moved did not affect the outcome. For observer locomotion parallel to the target, the change in target velocity that is predicted by the optic flow depends on the perceived distance of the target. Nevertheless, presenting the target at different distances (by presenting different images of the two eyes) did not affect the subjects' settings. The results show that our judgement of objects' velocities does not depend on an assessment of our own movements on the basis of a global analysis of the optic flow.