The aim of this study was to determine the minimum number of cardiac cycles necessary over which blood flow velocity representative of the situation (and not unduly influenced by sporadic irregularities). Blood flow was measured by Doppler ultrasound in the adult a. carotis communis, a. femoralis and fetal a. umbilicalis. The coefficients of variation (CV) for the Pulsatility Index (PI), Resistance Index (RI) and Time Average of Space Average Velocities (TASAV) from 10 averaged cycles were compared with the CVs of the respective indices from 2, 3, 4 up to 9 averaged cycles. The number of averaged cycles where the difference between the coefficients of variability became insignificant (Wilcoxon signed rank test) was taken to be the point at which representative values of PI, RI or, respectively, TASAV are assured. For PI evaluation for the a. carotis communis with either the 7 mx or 15 mx calculation program one needs to average over 7 cardiac cycles. Manually determining the maximum frequency curve envelope, 5 cycles are sufficient. Using these same calculation programs for the a. femoralis, 5, 6 and, respectively, 6, cycles are sufficient for representative PI indices; for the umbilical artery, 6, 8 and, respectively, 7 cycles are necessary. For RI, 5 cycles are sufficient in both the a. carotis communis and a. umbilicalis by manual evaluation. For TASAV, a minimum of 3 cycles for the a. carotis communis and of 7 for the a. femoralis are necessary. Thus the number of cardiac cycles which should be averaged is dependent on the vessel being examined, the index being calculated, and the program being used.