The American Thoracic Society (ATS) and the European Respiratory Society (ERS) recommend that the largest forced vital capacity (FVC) and the largest forced expiratory volume in 1 s (FEV1) should be recorded from at least three acceptable curves independently which curve they came from. Although these recommendations have been used for decades, there is still some controversy over their validity. The purpose of this study was to determine how the intersession variability of reported FVC and FEV1 values is influenced by different methods of selection in clinical practice. The study population consisted of 283 patients with obstructive airway diseases. Spirometry was performed until three acceptable forced expiratory curves were obtained in the standing position. A second set of spirometric measurements was obtained approximately 30 min after the first set of measurements. The following sampling methods were compared: method A, the largest FVC and the largest FEV1 among all three acceptable curves (ATS-ERS recommendation); method B, the FVC and the FEV1 from the single curve that yielded the largest sum of FVC plus FEV1 (best test); method C, the average of all three acceptable curves; method D, the average of the largest two FVCs and FEV1s among all of the three acceptable curves. FVC and FEV1 determined by method B gave almost identical values to those obtained by method A in most cases. However, method A was least variable for FEV1. In addition, the differences in FEV1 values between these two methods were large in some of patients with chronic obstructive pulmonary disease. The other selection criteria compared in this study offer no clear-cut advantages over method A. The ATS ERS recommended method appeared to be slightly more reproducible than the other selection criteria, including the 'best test' method, and should therefore be the preferred method of choice.