Lung function test can be influenced by technical as well as biological variables, and the quality control of test has been stressed as an important part of lung function test program. This study was done to survey the current status of lung function test program in workers' health examination, and examine the variablity of test results when quality control is not achieved. Forty nine Workers' Health Examination Institutes and 7 Pneumoconiosis Referal Centers were surveyed for their current lung function tests, types of spirometers, and performance of technicians. To examine the effect of quality control program of lung function tests on the variablity of test results, 118 miners were tested under a quality control guideline and the results were compared to those obtained ignoring 4 major areas of the guideline-spirometer calibration, minimum number of tests per session, suitable prediction equations, and selection process of final results for each tested subjects. Among surveyed Institutes and Centers, only 38%(20/53) are performing any calibration of spirometers, and most of those calibrations are done automatically by the machine. All the surveyed Institutes and Centers obtain acceptable test no more than 2 times. The background of technicians include office workers and nurse aids without an adequate training in pulmonary physiology, and the testing posture and formats vary among Institutes and Centers. When the types of pulmonary dysfunction under the definition of the Ministry of Labor were compared between those obtained with and without quality control guidelines, spirometer calibration affected the test results in 6(5.1%) out of 118 subjects. The change in the minimum number of adequate tests from 1 to 3 and more brought changes in 19(16.1%) subjects. The test results of 9(7.6%) subjects differed between two different prediction equations, and the results of 7(5.9%) subjects differed between two different selection process of final results. The current practice of lung function tests of workers needs quality control programs, and the results of this study indicate that the first remedy should be to bring the minimum number of acceptable test to more than 3.