Although current recommendations for spirometry require that the largest value of FEV1 and FVC should be taken from the largest values of different maneuvers, the validity of this approach was recently questioned. It has been suggested that selection of the maneuver with the largest peak flow or the maneuver with the largest FVC should be used for measurement of spirometric indices. The present analysis was therefore undertaken to determine which method of selection of spirometric maneuvers would give the least short-term variability in a clinical trial population. We examined the spirometry test sessions from 5,885 individuals with mild to moderate chronic airflow obstruction who were screened at two visits 24.9 +/- 17.1 d apart for entry into a multi-center clinical trial, the Lung Health Study. We compared eight potential selection methods for FEV1 and FVC. Using these different selection methods, the coefficient of variation ranged from 4.1 to 4.9% for FEV1 and from 3.5 to 5.7% for FVC. The average absolute difference between the two test sessions ranged from 110 to 123 ml for FEV1 and from 149 to 200 ml for FVC. Although all of the methods gave good results, the mean of the three highest values and the largest single value from all maneuvers provided the least short-term variability for both FEV1 and FVC. We therefore conclude that there is no reason to change the currently recommended selection methods for FEV1 and FVC.