Assessment of dynamic cerebral autoregulation (CA) requires continuous recording of arterial blood pressure (ABP). In humans, non-invasive ABP recordings with the Finapres device have often been used for this purpose. We compared estimates of dynamic CA derived from Finapres with those from invasive recordings in the aorta. Measurements of finger noninvasive-ABP (Finapres), intra-aortic-ABP (Millar catheter), surface ECG, transcutaneous CO₂ and bilateral cerebral blood flow velocity (CBFV) in the middle cerebral arteries (MCA) were simultaneously and continuously recorded in 27 patients scheduled for percutaneous coronary interventions. Phase, gain, coherence and CBFV step response from both the Finapres and intra-arterial catheter, were estimated by transfer function analysis. A dynamic autoregulation index (ARI) was also calculated. For both hemispheres, the ARI index and the CBFV step response recovery at 4s were significantly greater for the Finapres-derived estimates, than for the values obtained from aortic pressure. The transfer function gain for frequencies <0.1 Hz was significantly smaller for the Finapres estimates. The phase frequency response was significantly greater for the Finapres estimates at frequencies >0.1 Hz, but not at lower frequencies. The Finapres gives higher values for the efficiency of dynamic CA compared to values derived from aortic pressure measurements, as indicated by biases in the ARI index, CBFV step response, gain and phase. Despite the significance of these biases, their relatively small amplitude indicate a good level of agreement between indices of CA derived from the Finapres in comparison with corresponding estimates obtained from invasive measurements of aortic ABP.