The performance of a Vitalograph spirometer has been compared with a rolling seal spirometer (Ohio) to determine whether (i) spirometric measurements are normally distributed, (ii) fatigue occurs with repeated attempts, and (iii) how many tests are required. Twenty forced expiratory manoeuvres were performed on both spirometers at minute intervals, on ten normal subjects. To determine how many tests were required, the first five manoeuvres for each subject were analysed using eleven algorithms. The Vitalograph had a smaller volume, a shorter timing duration and a combined inertia and resistance greater than the European standards. The Ohio complied with the standards for volume and timing duration and had a much lower combined inertia and resistance. For each device, the spirometric indices were normally distributed, there were no fatigue effects, and no significant difference between any two algorithms. We conclude that (1) the performance of repeated forced expiratory manoeuvres using the Vitalograph spirometer does not result in fatigue, (2) spirometric indices are normally distributed, (3) estimation of forced expiratory flows between 25-50% of vital capacity from a Vitalograph may not be appropriate, and (4) the best of a number of technically satisfactory attempts, measured at one minute intervals, may be reported.