A group of 132 women and 12 men with goitre were studied to determine the prevalence of upper airway obstruction caused by the goitre. Inspection of flow-volume loops was used to detect upper airway obstruction and this suggested that 44 subjects (31 per cent) had the condition. Of these 44 subjects 19 per cent were men, which was a greater proportion than could be accounted for by chance. Flow-volume loops after surgery were recorded on 43 patients of whom 29 were from the group thought to have had upper airway obstruction. Comparison of measurements before and after surgery showed no important change in the 14 without, and improvement in 27 of those with, upper airway obstruction. Analysis indicated two failures of treatment and four probable false-positives among the group with upper airway obstruction. Inspection of the flow-volume loop had a 78 per cent specificity and 100 per cent sensitivity in detecting upper airway obstruction whereas an FEV:PEF ratio above 8 had a specificity of 94 per cent and a sensitivity of 64 per cent in this respect. Ultrasonography and plain radiography of the upper airway accurately predicted retrosternal extension of the goitre but did not predict upper airway obstruction. It is recommended that all patients with symptomatic goitre should have a flow volume loop recorded.