Continuous non-invasive blood pressure (CNBP) measurements were compared to invasive radial artery pressure recordings in 26 patients with cardiac, vascular and/or pulmonary disease. Patients were studied during general anaesthesia (n = 6), regional anaesthesia (n = 10), or combined technique (n = 10) for abdominal or transurethral surgery. CNBP was obtained from a cuff placed around the upper arm and simultaneously compared to invasive pressure from the ipsilateral radial artery. A CNBP device (7001 Cortronic) used intermittent oscillometric measurement for calibration. Through a cuff continuously inflated to a pressure of 20 mmHg, a microprocessor-controlled electro-pneumatic acquisition system sensed displacements of the brachial artery wall. Amplified, digitally converted, filtered and transformed data were displayed as a continuous pulse pressure waveform and digital pressure values on the screen. The CNBP method functioned without disturbances before surgery in all patients. Intra-operative use of electrocautery or a spontaneous occurrence of warning on the screen repeatedly triggered oscillometric recalibration, hence CNBP measurements were discontinued in nine patients. Coefficients of correlation (r) of all invasive and CNBP pairs (n = 1111) were 0.68, 0.58 and 0.70 for systolic, diastolic, and mean blood pressures, respectively. Prediction errors (bias, mean +/- SD) were -13.6 +/- 22.5 mmHg (on average CNBP < invasive pressure) for systolic, +13.0 +/- 12.4 mmHg (CNBP > invasive pressure) for diastolic and +5.0 +/- 13.9 mmHg (CNBP > invasive pressure) for mean CNBP, as compared to radial artery pressure values. Absolute errors (precision) were 25.3 +/- 9.4 mmHg for systolic, 17.4 +/- 4.5 mmHg for diastolic, and 13.9 +/- 4.6 mmHg for mean CNBP. During anaesthesia induction (n = 672) the difference between consecutive measurements (trend of pressure changes) with invasive and CNBP method exceeded 20 mmHg in 90 (13.3%) instances for systolic, in 33 (4.9%) instances for diastolic, and in 45 (6.6%) instances for mean blood pressure. In conclusion, the CNBP method by brachial artery wall displacement failed to measure the blood pressure reliably and to display the trend of pressure changes correctly during anaesthesia induction. In its present form this CNBP method should not replace invasive blood pressure monitoring in high-risk patients neither for anaesthesia induction nor during non-thoracic surgical procedures.