Background. Nitrates and furosemide, commonly administered in the treatment of pulmonary oedema, have not been compared in a prospective clinical trial. We compared the efficacy and safety of these drugs in a randomised trial of patients with severe pulmonary oedema and oxygen saturation below 90%. Methods. Patients presenting to mobile emergency units with signs of congestive heart failure were treated with oxygen 10 L/min, intravenous furosemide 40 mg, and morphine 3 mg bolus. 110 patients were randomly assigned either to group A, who received isosorbide dinitrate (3 mg bolus administered intravenously every 5 min; n = 56) or to group B, who received furosemide (80 mg bolus administered intravenously every 15 min, as well as isosorbide dinitrate 1 mg/h, increased every 10 min by 1 mg/h; n = 54). Six patients were withdrawn on the basis of chest radiography results. Treatment was continued until oxygen saturation was above 96% or mean arterial blood pressure had decreased by 30% or to below 90 mmHg. The main endpoints were death, need for mechanical ventilation, and myocardial infarction. The analyses were by intention to treat. Findings. Mechanical ventilation was required in seven (13%) of 52 group-A patients and 21 (40%) of 52 group-B patients (p = 0.0041). Myocardial infarction occurred in nine (17%) and 19 (37%) patients, respectively (p = 0.047). One patient in group A and three in group B died (p = 0.61). One or more of these endpoints occurred in 13 (25%) and 24 (46%) patients, respectively (p = 0.041). Interpretation. High-dose isosorbide dinitrate, given as repeated intravenous boluses after low-dose intravenous furosemide, is safe and effective in controlling severe pulmonary oedema. This treatment regimen is more effective than high-dose furosemide with low-dose isosorbide nitrate in terms of need for mechanical ventilation and frequency of myocardial infarction.