A comparison was made of automated versus manual measurement of pulmonary artery (PA) and wedge (WP) pressures. The manual pressure measurements were taken at end-expiration whereas the automated measurements were taken using existing monitor and computer algorithms. A total of 40 critical care patients were divided into groups according to the ventilatory mode used (spontaneous, intermittent mandatory ventilation [IMV], or assist/control). In patients who were breathing spontaneously, the automated method underestimated mean PA pressure (MPAP) (p less than 0.01), WP (p less than 0.001), and PA diastolic (p less than 0.001) pressure but not PA systolic pressure. In patients on IMV, the automated method underestimated MPAP (p less than 0.05), WP (p less than 0.001), and PA diastolic (p less than 0.001) pressure and overestimated PA systolic pressure (p less than 0.05). In patients on assist/control, the automated method overestimated WP (p less than 0.001) and PA systolic (p less than 0.005) pressure pressure, underestimated PA diastolic (p less than 0.001) pressure and did not affect MPAP. The error was not affected by respiratory rate, thoracic compliance, or level of PEEP. The errors in automated pressure measurements believed to be clinically important varied with the ventilatory mode used. Patients breathing spontaneously had the largest measurement error, with 42% of these patients having a clinically important error in WP and 99% having a clinically important error in PA diastolic pressure. Patients on assist/control had the fewest errors in automated pressure measurements. In all ventilatory modes used, automated measurement of PA diastolic pressure had the largest amount of error.