The dynamic relationship between spontaneous fluctuations of arterial blood pressure (ABP) and corresponding changes in cerebral blood flow velocity (CBFV) is studied in a population of 83 neonates. Static and dynamic methods are used to identify two subgroups showing either normal (group A, n = 23) or impaired (group B, n = 21) cerebral autoregulation. An FFT algorithm is used to estimate the coherence and transfer function between CBFV and ABP. The significance of the linear dependence between these two variables is demonstrated by mean values of squared coherence > 0.50 for both groups in the frequency range 0.02-0.50 Hz. However, group A has significantly smaller coherences than group B in the frequency ranges 0.02-0.10 Hz and 0.33-0.49 Hz. The phase response of group A is also significantly more positive than that of group B, with slopes of 9.3 +/- 1.05, and 1.80 +/- 1.2 rad Hz-1, respectively. The amplitude frequency response is also significantly smaller for group A in relation to group B for the frequency range 0.25-0.43 Hz. These results suggest that transfer function analysis may be able to identify different components of cerebral autoregulation and also provide a deeper understanding of recent findings by other investigators.