A continuous, noninvasive device for blood pressure measurement using pulse transit time has been recently introduced. We compared blood pressure measurement determined using this device with simultaneous invasive blood pressure measurements in 35 patients undergoing general endotracheal anesthesia. Data were analyzed for accuracy and tracking ability of the noninvasive technique, and for frequency of unavailable pressure measurements by each method. A total of 25,133 measurements of systolic pressure, diastolic pressure, and mean arterial pressure (MAP) by each method were collected for comparison from 35 patients. Accuracy was expressed by reporting mean bias (invasive pressure minus noninvasive pressure) and limits of agreement between the two measurements. After correction for the offset found when measuring invasive and oscillometric methods of arterial pressure measurement, the mean biases for systolic, diastolic, and mean pressures by the pulse wave method were -0.37 mm Hg, -0.01 mm Hg, and -0.05 mm Hg, respectively (p < 0.001). The limits of agreement were: -29.0 to 28.2 mm Hg, -14.9 to 14.8 mm Hg, and -19.1 to 19.0 mm Hg, respectively (95% confidence intervals). When blood pressure measured invasively changed over time by more than 10 mm Hg, the noninvasive technique accurately tracked the direction of change 67% of the time. During the entire study, 3.2% of the invasive measurements were unavailable and 12.9% of the noninvasive measurements were unavailable. The continuous noninvasive monitoring technique is not of sufficient accuracy to replace direct invasive measurement of arterial blood pressure, owing to relatively wide limits of agreement between the two methods.(ABSTRACT TRUNCATED AT 250 WORDS)