Eighteen patients with moderate to severe pulmonary hypertension were studied, nine with intracardiac shunts and nine without. The effects of an incremental infusion of epoprostenol (prostacyclin) (0.5-8 ng/kg per minute) or sublingual nifedipine (20-30 mg) were compared with the response to three months' treatment with oral nifedipine. Both epoprostenol and sublingual nifedipine caused a fall in pulmonary vascular resistance and pressure and a rise in cardiac output. Patients with intracardiac shunts had higher systemic blood flows than those without shunts. Exercise in the shunt group was accompanied by systemic desaturation and hyperventilation. Analysis of individual results showed that the size of the response was inversely related to the severity of the pulmonary vascular disease. A good long term response to nifedipine seemed to be as readily predicted by the resting control values for haemodynamic variables as by values after short term treatment. A favourable response was likely if the pretreatment mean pulmonary artery pressure was less than 50 mm Hg, the ratio of total pulmonary to systemic resistance was less than 0.7, or the ratio of mean pulmonary artery pressure to systemic artery pressure was less than 0.6. Short term vasodilator protocols may do harm. If such studies are carried out, an adequate dose range must be tried before the long term efficacy of an individual drug can be forecast.