The blood in the carotid arteries exhibits time-varying flow velocity as a function of cardiac phases. Despite this flow velocity variation, most current methods set forth for the analysis of arterial spin labeling (ASL) data have assumed that the tagged blood is delivered from the tagging region to the imaging region via simple plug flow, i.e., a single transit delay (deltat). In this study, we used a pulse oximeter to synchronize image acquisition at systole and diastole separately. The deltat dispersion was modeled with a Gaussian distribution and the effect of cardiac pulsation upon the ASL signal was evaluated on five healthy volunteers. ASL signals were collected at a series of inflow times (TI) using PICORE QUIPSS II: TR/TE/TI1 = 2400/3.2/700 ms, TI = {300, 500, 700, 900, 1100, 1300, 1500} ms, matrix size = 64 x 64, repetition = 100. Transit delay was found significantly shorter in systolic tag than diastolic tag (paired student's t-test, p < 0.001; mean difference across subjects = 54 ms). When the tag was applied in late systole, the ASL signal arrived in the target brain slice earlier, and was higher by 16% with TI = 700 ms. Intervoxel dispersion (-350 ms) dominated over intravoxel dispersion (< 200 ms). The disparity of ASL signals found between systolic and diastolic tags indicated that ASL imaging was sensitive to cardiac pulsations. We conclude that both flow dispersion and fluctuations in the ASL signal due to cardiac pulsations are significant.