This study aims to compare automatic oscillometric blood pressure recordings with simultaneous direct intra-arterial blood pressure measurements in hyperacute stroke patients to test the accuracy of oscillometric readings. A total of 51 first-ever stroke patients underwent simultaneous noninvasive automatic oscillometric and intra-arterial blood pressure monitoring within 3 h of ictus. Casual blood pressure was measured in both arms using a standard mercury sphygmomanometer on hospital admission. Patients who received antihypertensive medication during the blood pressure monitoring were excluded. The estimation of systolic blood pressure (SBP) using oscillometric recordings underestimated direct radial artery SBP by 9.7 mmHg (95% confidence interval: 6.5-13.0, P<0.001). In contrast, an upward bias of 5.6 mmHg (95% confidence interval: 3.5-7.7, P<0.001) was documented when noninvasive diastolic blood pressure (DBP) recordings were compared with intra-arterial DBP recordings. For SBP and DBP, the Pearson correlation coefficients between noninvasive and intra-arterial recordings were 0.854 and 0.832, respectively. When the study population was stratified according to SBP bands (group A: SBP<or=160 mmHg; group B: SBP>160 mmHg and SBP<or=180 mmHg, group C: SBP>180 mmHg), higher mean DeltaSBP (intra-arterial SBP-oscillometric SBP) levels were documented in group C (+19.8 mmHg, 95% confidence intervals: 12.2-27.4) when compared with groups B (+8.5 mmHg, 95% confidence intervals: 2.7-14.5; P=0.025) and A (+5.9 mmHg, 95% confidence intervals: 1.8-9.9; P=0.002). Noninvasive automatic oscillometric BP measurements underestimate direct SBP recordings and overestimate direct DBP readings in acute stroke. The magnitude of the discrepancy between intra-arterial and oscillometric SBP recordings is even more prominent in patients with critically elevated SBP levels.