The distance-calibration hypothesis states that retinal velocity is scaled by using distance cues, and judged velocity remains unchanged when distance is changed. The relational hypothesis states that judged velocity depends on retinal velocities, and is proportional to judged distance. These hypotheses were compared in three experiments where the movements of the standard stimulus and the comparison stimulus were manipulated by the ratio of the angular velocity of the comparison stimulus to the angular velocity of the standard stimulus. The presentation conditions of the standard stimulus and the comparison stimulus, and the colour cues of the two stimuli were also manipulated in order to change the strength of the cues available to the observers. The results indicate that judged velocities and the relationship of judged distance and velocity depend on the strength of the cues. When cues are strong, the distance-calibration hypothesis adequately explains speed constancy. When cues are weak, judged velocity and the relationship between judged distance and velocity are consistent with the prediction of the relational hypothesis. The perceived speed of a stimulus depends not only on the physical speed of the stimulus but also on non-motion cues, some of which are distance cues involved in depth perception.