Series of single-shot blipped echo-planar images with spin-density weighting and T2* sensitivity (2.0 T, TR = 400 ms, TE = 54 ms, flip angle = 30 degrees) were used to study the temporal response profiles to repetitive visual activation (5 Hz, reversing black and white checkerboard versus darkness) for protocols comprising multiple cycles of a 1.6-s stimulus in conjunction with a 8.4-s or 90-s recovery phase and a 10-s stimulus with a 20-s or 90-s recovery phase. Analysis of the real-time data from all activated pixels resulted in a strong positive MRI response (mean values 3-6%) as well as a marked poststimulus undershoot (mean values 1-2%, duration 60-90 s) for all paradigms. Repetitive protocols with insufficient recovery periods of 8.4 s or 20 s gave rise to a wraparound effect when analyzing time-locked averages from multiple activation cycles. This problem may lead to an early signal decrease that originates from the ongoing undershoot of preceding activations folded back into the initial latency phase of a subsequent activation. When ensuring complete decoupling of responses to successive stimuli by using a 90-s recovery period, the wraparound effect vanished and an initial dip was observed in one of seven subjects for a 10-s/90-s protocol.