We studied visual perception and gaze control in nine participants while they judged the relative phase between two oscillating stimuli (Experiment 1), and while they moved their hand - and therewith a concurrent feedback signal - in-phase or in antiphase with an oscillating stimulus (Experiment 2). As in previous studies, the mean relative phase judgements in Experiment 1 corresponded to the presented phase relations (0°, 45°, 90°, 135°, and 180°), whereas their standard deviations followed an inverted U-function of relative phase. The relative phase judgements were hardly affected by the degree of visibility (fully visible, inner parts occluded, outer parts occluded) and the amplitude (5°, 10°, and 20°) of the stimuli. Stimulus-gaze coupling decreased as relative phase increased, and its variability correlated with that of the relative phase judgements. Taken together, task performance and gaze behaviour suggested that the judgement of relative phase might be flexibly based on different variables, rather than a single variable like relative direction of motion. In Experiment 2, the production of the antiphase relation was less stable than that of the in-phase relation. Performance deteriorated when the outer parts of the signals were occluded and when their amplitudes were reduced. Stimulus-gaze coupling was stronger during in-phase than during antiphase tracking and weaker when the signals were partially occluded and when their amplitudes were reduced. Stimulus-gaze coupling at 0° and 180° was stronger in Experiment 2 than in Experiment 1, suggesting that the visual perception of relative phase may benefit from its active production. Overall, the results clearly indicated that visual perception of relative phase and the corresponding gaze control are strongly task-dependent. © 2005 Elsevier B.V. All rights reserved.