The accuracy of 24-h blood pressure values obtained by ambulatory monitoring via the Spacelabs 5300 device was evaluated by comparison with simultaneous 24-h intra-arterial blood pressure recording from the contralateral arm. The comparison was made in eight essential hypertensive subjects in whom non-invasive blood pressure was measured every 15 (day) or 30 min (night). The measurements were automatically and visually edited to eliminate artefactual readings and hourly and 24-h means were calculated separately for systolic and diastolic blood pressure. The corresponding intra-arterial blood pressure means were also calculated. In the group as a whole, hourly means obtained by the non-invasive device were similar or only slightly different from those recorded intra-arterially. The 24-h systolic blood pressure mean obtained non-invasively was not significantly different from that obtained intra-arterially (138.4 +/- 9.1 and 142.9 +/- 9.2 mmHg, respectively), nor were the corresponding 24-h diastolic blood pressure means significantly different (83.5 +/- 4.5 and 80.6 +/- 3.5 mmHg, respectively). However, in spite of these similarities, there were contrasting and often large discrepancies between non-invasive and intra-arterial values in individual subjects. For the 24-h systolic blood pressure mean the discrepancies ranged from 7.6 +/- 1.1 to 16.1 +/- 2.2 mmHg and for the 24-h diastolic blood pressure mean, from 3.5 to 13.2 mmHg. Thus, the Spacelabs 5300 device has a limited ability to correctly estimate ambulatory blood pressure in individual subjects. It may be better suited for the estimation of group blood pressures, but only because errors are smoothed by the summation of individual errors of opposing signs.