In order to assess the accuracy of spectral analysis of beat-by-beat arterial blood pressure fluctuations recorded by the finger technique we analysed simultaneous ipsilateral resting sequences (180 s) of systolic blood pressure (SP) and diastolic blood pressure (DP) obtained by brachial artery cannulation and by the Finapres device in 26 post-myocardial infarction (post-MI) and 24 chronic heart failure (CHF) patients. We estimated: (i) the total variability and the power in the bands 0.01-0.04 Hz (band 1), 0.04-0.07 Hz (band I), 0.07-0.15 Hz (band III) and 0.15-0.45 Hz (band IV); and (ii) the group-average transfer function between the invasive and non-invasive channel. The total variability of intra-arterial SP was greatly enhanced by the Finapres device in both populations (p < 0.01) mainly due to the increase of power in bands I and II (mean percentage errors 157 and 111% in post-MI and 368 and 245% in CHF patients respectively). The power in band IV was significantly reduced in post-MI (p < 0.001) but not in CHF patients. The total DP variability increased in post-MI patients (p < 0.001) due to the increase in bands I and II (mean percentage errors 62 and 43% respectively). No significant difference was found in bands III and IV of the same group and in the overall variability of CHF. Linear system analysis showed that intra-arterial SP oscillations in both groups are amplified by the Finapres device in bands I and II, with a higher gain in CHF patients, whereas they are attenuated in band IV. Band III acts as a transition band. DP oscillations recorded non-invasively are slightly amplified in bands I and II of post-MI patients only and are attenuated in band IV of both groups. A substantial fraction (approximately 30%) of the non-invasive SP variability in band I cannot be explained as a pure linear transformation (amplification) of the invasive signal. A phase delay in the range 7 degrees-10 degrees relates the non-invasive signal to the invasive one in the region 0.025-0.17 Hz and approaches zero elsewhere. Several mechanisms were examined as potential causes of the invasive-non-invasive differences. We argue that these differences are mainly due to the complex rhythmic behaviour of peripheral resistance vessels.