The spatial and temporal parameters determining the duration of the jitter aftereffect (JAE) were examined. Experiment one showed the JAE is a luminance-based effect. Experiment two showed that element sizes 0.084 square (≈7 cpd) and temporal frequencies above 18 Hz were optimal. Experiment three showed that the JAE is dependent on the rate of change during adaptation, not the number of changes. Experiment four compared directional noise: linear, circular, and radial to adirectional dynamic random noise (DRN). Linear noise was better than circular or radial, but random noise was best. Experiment five tested horizontal and vertical oscillation of various peak-to-peak amplitudes, frequencies and velocities. Velocity had a significant effect on JAE duration; amplitude did not. Murakami and Cavanagh’s (1998) proposal that the region of least instantaneous motion becomes the new baseline for perceived zero motion was tested. A motion energy difference between regions is necessary for the JAE. A difference in motion directions between regions at the same energy level is not sufficient. The JAE required one region above 18 Hz, and one below it. Experiment six compared Brownian motion to DRN of identical energy levels. DRN always produced a stronger JAE. Contrast was tested and found to be effective only during adaptation. A dynamic theory where miniature eye movements facilitate relative motion perception was proposed to account for the JAE.