The variability of the valve gradient measured by Doppler in pulmonary stenosis was compared with the variability of the gradient measured at catheterisation in 42 infants and children undergoing catheterisation with a view to balloon dilatation of the pulmonary valve. The maximum value measured by Doppler when the patient was unsedated was significantly higher than that measured when the patient was sedated for catheterisation, and the maximum gradient was significantly higher shortly after than several days later. In a patient with pronounced infundibular obstruction after dilatation the Doppler signal clearly showed that the obstruction was dynamic, with a superimposed lower fixed signal that correctly predicted the final low gradient. The Doppler gradient in an alert and unsedated patient may be a better measure of the true physiological value. The highest Doppler value so obtained is a more appropriate indicator of the need for balloon dilatation than a single catheter measurement. The result of dilatation is best assessed by Doppler measurement at least a day after the procedure.