Many physiological signals fluctuate in an apparently irregular and complex manner. The fractal analysis of these changes often confirms the existence of long-range correlations in healthy subjects and demonstrate a lack of such correlations in pathological conditions. The authors discuss usefulness of the fractal statistics methods in the analysis of intracranial blood flow velocity changes averaged over a cardiac beat interval measured in humans, using transcranial Doppler ultrasonography (TCD). An assumption was made that fractal properties of blood velocity time series in healthy individuals result from a proper autoregulation of their cerebral blood flow and that in cases of vascular disturbances fractuality may disappear. A review of the literature is presented to provide a theoretical rationale for the proposed method.