If intravenous access is not available during cardiopulmonary resuscitation, tracheal administration of adrenaline 0.02 mg/kg, twice the intravenous dose, is recommended. In a randomised crossover study we investigated the haemodynamic effects of low doses of tracheal versus intravenous adrenaline. 12 anaesthetised patients having a hip replaced received one dose of adrenaline intravenously (0.1 microgram/kg) and the other tracheally (0.5 microgram/kg). There was a mean increase in systolic arterial pressure of 40.5 mm Hg (range 16-81) after the intravenous injection, with little effect on heart rate. Tracheal adrenaline had no effect on arterial pressure or heart rate. Thus low doses of tracheal adrenaline have no haemodynamic effects. We believe that the recommended tracheal dose of twice the intravenous dose is likely to be ineffective for the treatment of cardiac arrest. Animal studies suggest that a tracheal dose at least ten times the intravenous dose is required.