The present study compares the spectral characteristics of 24-h blood pressure variability estimated invasively at the brachial artery level with those estimated by measurement of blood pressure at the finger artery using the non-invasive Portapres device. Broad-band spectra (from 3x10(-5) to 0.5 Hz) were derived from both finger and intra-brachial pressures recorded simultaneously for 24 h in eight normotensive and twelve hypertensive ambulant subjects. At frequencies lower than 0.07 Hz, higher spectral estimates were obtained by Portapres than by intra-brachial measurements. The maximum overestimation occurred in systolic pressure at around 10(-2) Hz, where the amplitude of the oscillations was two times greater when measured by Portapres. A less pronounced overestimation was found for diastolic pressures. The maximum overestimation was greater during daytime than during night-time. At around 0.1 Hz, invasive and non-invasive spectra were similar. At the respiratory frequencies (0.15-0.50 Hz), the power spectra were overestimated by Portapres during daytime, and underestimated at night. These results provide reference information for the correct interpretation of Portapres data in the estimation of 24-h blood pressure spectral power.