We have used the technique of positron emission tomography to study and compare the cortical activity produced when humans view a pattern of small squares moving incoherently with respect to one another and when the same pattern moves coherently and unidirectionally. A stationary version of the stimulus acted as a control. Our choice of paradigm was inspired by psychophysical models and physiological studies in the macaque monkey which show that directionally selective cells in V5 respond optimally to unidirectional coherent motion, whereas those of V1 respond to motion within their receptive fields, regardless of the motion in surrounding parts. Our results show that human V1/V2, V3, and V5 are all activated by both types of motion stimuli. Incoherent motion, however, proved to be more effective than coherent motion in activating V1/V2 and V5. Thus the higher perceptual salience of unidirectional coherent motion in comparison to incoherent motion is not reflected by any increased activation of human area V5.