Balloon dilatation of the aortic valve was attempted 16 times in 15 patients with severe aortic stenosis. None died but one had a transient stroke after the procedure. At dilatation the gradient across the aortic valve was reduced by greater than 30% in 69% of patients and the Gorlin valve area (calculated in 7/15 patients) increased by 30% in half. But a comparison of Doppler gradients measured before and one to two days after dilatation in 11 patients showed a greater than 30% reduction in the simultaneously measured gradient in only four. Doppler gradient was the most accurate predictor of symptomatic benefit and a fall in Doppler gradient persisted mainly in patients whose peak to peak gradient fell by at least 40% at the time of the procedure. Balloon dilatation of the aortic valve is a relatively safe procedure but it is less successful than previous reports suggest, perhaps because of early restenosis. Some forms of aortic stenosis may be more amenable to this procedure than others.