An ultrasonic flowmeter could be advantageous over a differential pressure pneumotachograph having a constant error in varying conditions. The in vitro accuracy of ultrasonic tidal volume (VT) estimates for ventilated infants were evaluated. Flow linearity and frequency response were tested, as was the influence of humidity and oxygen content on the accuracy of VT estimates. The linearity was within the 5% limits between -350 and 350 mL x sec(-1) and was not affected by the presence of an endotracheal tube (ET). The frequency response was flat and unaffected by an ET up to 4.5 Hz. The VT in the range 7-100 mL, in air showed a mean error of 0.1% (95% confidence interval (CI) -0.2-0.4%) with a maximum and minimum of 6.5 and -3.5% respectively. Humidity did not affect accuracy. After calibration in air, the maximal mean error for measurements in pure oxygen was 3.0% (95% CI 1.9-4.1%). Repeated measurements over 5.5 h had a mean error of 0.4% (95% CI -0.7-0.1%). The in vitro evaluation of an ultrasonic flowmeter showed stable accuracy in mechanical ventilation conditions. Changing connection geometry and oxygen content did not increase the error to a clinically relevant degree. The flowmeter could therefore be a better alternative than the pneumotachograph for ventilated infants.