Objective measurement of bronchial damage in patients with bronchiectasis is needed to identify progressive disease. This study evaluates the inter- and intraobserver variation in computed tomography (CT) measurements of bronchial wall circumference and examines the precision with which CT measurements of the bronchi can be repeated in patients with bronchiectasis. Twelve patients were scanned and the circumferences of 61 subsegmental bronchi were measured independently on two occasions, by three observers, using a tracing facility on the CT console. To determine the accuracy with which previously acquired sections could be repeated, five patients were scanned on two separate occasions. The mean bronchial circumference measured by the three observers was 16.1 +/- 8.3 mm. The standard deviations of differences between first and second measurements of bronchial circumference for the three observers (intraobserver variability) were: 0.60 mm, 0.67 mm and 0.40 mm. The standard deviation of readings for interobserver variability was 0.71 mm. The standard deviations of differences (and coefficients of variation) for measurements of bronchial circumference following rescanning were: 1.82 mm, 1.40 mm and 1.74 mm (9.9%, 7.6% and 9.3%, respectively). The reproducibility of measurements of wall circumference, between and within observers and between examinations, indicates that such measurements may be clinically useful in demonstrating the progression of bronchiectasis.