OBJECTIVE To evaluate the accuracy and reliability of noninvasive continuous finger blood pressure measurement in critically ill patients. METHODS Prospective data collection. METHODS Emergency department in a 2,000-bed hospital. METHODS Thirty-nine patients admitted to the emergency department requiring invasive arterial blood pressure monitoring were enrolled to the study protocol. METHODS Continuous noninvasive blood pressure measurement was performed on the middle phalanx of the second and third finger, using a test instrument which provides continuous arterial waveform display with the use of a finger cuff. Invasive mean arterial blood pressure measurement was done by cannulation of the radial artery and direct transduction of the systemic arterial pressure waveform. RESULTS Three thousand one hundred eighteen pairs of simultaneous finger cuff and intra-arterial blood pressure measurements were collected in 1-min intervals from 39 patients over a total of 51.8 hrs. The overall discrepancy between both measurements was 0.10 mm Hg. The standard deviation of the differences was +/- 5.02 mm Hg. The mean bias in patients treated with catecholamines was 0.01 mm Hg and was not different from the bias observed in patients without catecholamines (mean bias: 0.23 mm Hg; p > .22). Whereas 95% of all comparisons between finger cuff and intra-arterial measurement had a discrepancy < or = +/- 10 mm Hg, 4.7% had a discrepancy between +/- 10.1 to 15 mm Hg and 0.3% exhibited a discrepancy > +/- 15 mm Hg. In 29 (74%) patients, the duration of errors > 10 mm Hg was < or = 1 min. In seven (18%) patients, the duration of errors > 10 mm Hg was between 2 to 3 mins and in three (8%) patients, the errors lasted for > 3 mins. CONCLUSIONS Our data provide a guide to the accuracy and reliability of noninvasive finger blood pressure measurements in critically ill patients. Although most test instrument measurements were reliable, in 8% of all patients large discrepancies (> 10 mm Hg) between both measurements with a duration of > 3 mins were noted. Concerning the considerable risk for arterial cannulation, our preliminary data demonstrate that the test instrument (PORTAPRES, TNO Biomedical Instrumentation Research Unit; The Netherlands) is an advance in noninvasive monitoring of critically ill patients and may be useful in most emergency clinical settings.