Summary The reproducibility of ventilatory parameters of routine spirometry was statistically evaluated on three different groups of male adults, composed of 6 well-trained doctors, 30 healthy medical students, and 12 cooperative chest patients.The normal frequency distribution for VC, FEV1.0, and FEV1 .0% was established with 600 spirometric measurements on the 6 doctors, in which these spirometric values were represented as percent deviation of the mean calculated on each subject. The variations of VC, FEVI .0, and FEV 1.0 % were consistently smaller than those of MMF (FEF25-75%) and MVV.Confidence intervals of 95 % for the mean of VC, FEV1.0, FEV1.0%, MMF, and MVV were presented on each of the groups. Most 95 confidence intervals on the groups of doctors and students seemed to be smaller than those on the group of patients. Statistical analysis of the difference in 95 % confidence interval was described.The authors recommended the representation of spirometric parameters by a mean estimated from at least three successful spirometries.Routine spirometry is a widely used procedure which provides us information for the assessment of overall ventilatory capacity. Appreciable variations of the data obtained by a subject are often noticed on repeated spirometry, even though the tests are conducted satisfactorily. Since the numbers of tests to be repeated on a subject are limited in routine practice, effective approach is essential in estimating the magnitude of variation of the data in terms of the numbers of repeated tests . Precise information on qualitative and quantitative variations of the data obtained by a subject on the repeated tests must be made available for assessment of ventilatory capacity. In the present experimental observations, the reproducibility of spirometric parameters is statistically evaluated on the groups of healthy