Although transcranial Doppler ultrasound (TCD) has been used to detect oscillations in CBF, interpretation is severely limited, since only blood velocity and not flow is measured. Oscillations in vessel diameter could, therefore, mask or alter the detection of those in flow by TCD velocities. In this report, the authors use a TCD-derived index of flow to detect and quantify oscillations of CBF in humans at rest. A flow index (FI) was calculated from TCD spectra by averaging the intensity weighted mean in a beat-by-beat manner over 10 seconds. Both FI and TCD velocity were measured in 16 studies of eight normal subjects at rest every 10 seconds for 20 minutes. End tidal CO2 and blood pressure were obtained simultaneously in six of these studies. The TCD probe position was meticulously held constant. An index of vessel area was calculated by dividing FI by velocity. Spectral estimations were obtained using the Welch method. Spectral peaks were defined as peaks greater than 2 dB above background. The frequencies and magnitudes of spectral peaks of FI, velocity, blood pressure, and CO2 were compared with t tests. The Kolmogorov-Smirnov test was used to further confirm that the data were not white noise. In most cases, three spectral peaks (a, b, c) could be identified, corresponding to periods of 208+/-93, 59+/-31, and 28+/-4 (SD) seconds for FI, and 196+/-83, 57+/-20, and 28+/-6, (SD) seconds for velocity. The magnitudes of the spectral peaks for FI were significantly greater (P<0.02) than those for velocity. These magnitudes corresponded to variations of at least 15.6%, 9.8%, and 6.8% for FI, and 4.8%, 4.2%, and 2.8% for velocity. The frequencies of the spectral peaks of CO2 were similar to those of FI with periods of 213+/-100, 60+/-46, and 28+/-3.6 (SD) seconds. However, the CO2 spectral peak magnitudes were small, with an estimated maximal effect on CBF of (+/-) 2.5+/-0.98, 1.5+/-0.54, and 1.1+/-0.31 (SD) percent. The frequencies of the blood pressure spectral peaks also were similar, with periods of 173+/-81, 44+/-8, and 26+/-2.5 (SD) seconds. Their magnitudes were small, corresponding to variations in blood pressure of (+/-) 2.1+/-0.55, 0.97+/-0.25, and 0.72+/-0.19 (SD) percent. Furthermore, coherence analysis showed no correlation between CO2 and FI, and only weak correlations at isolated frequencies between CO2 and velocity, blood pressure and velocity, or blood pressure and FI. The Kolmogorov-Smirnov test distinguished our data from white noise in most cases. Oscillations in vessel flow occur with significant magnitude at three distinct frequencies in normal subjects at rest and can be detected with a TCD-derived index. The presence of oscillations in blood velocity at similar frequencies but at lower magnitudes suggests that the vessel diameters oscillate in synchrony with flow. Observed variations in CO2 and blood pressure do not explain the flow oscillations. Ordinary TCD velocities severely underestimate these oscillations and so are not appropriate when small changes in flow are to be measured.