Estimates of cerebral critical closing pressure (CrCP) and resistance-area product (RAP) are often derived using noninvasive measurements of arterial blood pressure (ABP) in the finger, but the errors introduced by this approach, in relation to intra-vascular measurements of ABP, are not known. Continuous recordings of ABP (Finapres and solid-state catheter-tip transducer in the ascending aorta), cerebral blood flow velocity (CBFV, bilateral Doppler), ECG and transcutaneous CO(2) were performed following coronary catheterization. CrCP and RAP were calculated for each of 12,784 cardiac cycles from 27 subjects using the classical linear regression (LR) of the instantaneous CBFV-ABP relationship and also the first harmonic (H(1)) of the Fourier transform. There was a better agreement between LR and H(1) for the aortic measurements than for the Finapres (p < 0.000,01). For LR there were no significant differences for either CrCP or RAP due to the source of ABP measurement, but for H(1) the differences were highly significant (p < 0.000,03). The coherence functions between either CrCP or RAP values calculated with aortic pressure (input) or the Finapres (output) were significantly higher for H(1) than for LR for most harmonics below 0.2 Hz. When using the Finapres to estimate CrCP and RAP values, the LR method produces similar results to intra-arterial measurements of ABP for time-averaged values, but H(1) should be preferred in applications analysing beat-to-beat changes in these parameters.