We compared values of invasive blood pressure measured intra-arterially with those measured noninvasively with an automated oscillometric monitor. Twenty-eight patients undergoing cardiac surgical procedures under general anesthesia were studied and 552 determinations were made. The two methods of measuring blood pressure correlated within the expected bounds of experimental accuracy and physiological variation. However, the correlation between invasive and noninvasive methods varied, apparently arbitrarily, with time. These disparities could not be explained by a linear combination of physiological variables recorded. Systolic determinations correlated the best and diastolic the least between the invasive and noninvasive methods. In general, the correlation was better for adults than for children, except with diastolic blood pressure.