This study determined which variables affected endotracheal tube "leak" pressures in 80 surgical patients, two weeks to 11 years of age, intubated with uncuffed tracheal tubes. We defined "leak" pressure as the inspiratory pressure needed to cause an audible escape of gas around the endotracheal tube. "Leak" pressure was measured after varying either head position, tracheal tube depth within the trachea, fresh gas flow rate, or degree of neuromuscular block. "Leak" pressure increased progressively from 16.9 +/- 1.3 cmH2O with complete patient paralysis to 30.6 +/- 1.4 cmH2O following 100 per cent recovery of neuromuscular function. Turning the head from a neutral position to one side increased "leak" pressure from 14.7 +/- 1.7 cmH2O to 24.4 +/- 2.5 cmH2O. Varying tracheal tube depth or fresh gas flow rate produced no significant change in "leak" pressure. Thus, there may be marked variability in "leak" pressure, depending on head position and degree of neuromuscular blockade. Keeping the patient fully paralyzed with the head in a neutral position provides a reliable and consistent method for measuring "leak" pressures.