Lung function testing of a random population sample in the eight SAPALDIA (Swiss study on air pollution and lung diseases in adults) centres had to be performed simultaneously, within one year, by eight teams and 23 technicians. We conducted quality control studies to test for technician, team and device related systematic measurement errors. To assess technician effects, each centre conducted a study involving 12-19 subjects. Two studies with 13 participants each addressed team and device effects. In all studies, volunteers repeatedly performed spirometry with different technicians or devices. Effects due to technician, team or device were estimated (analysis of variance). Neither "technician" within any of eight teams nor "team" accounted for significant differences of forced vital capacity (FVC) or forced expiratory volume in one second (FEV1). The Device Effect Study revealed 10% lower FVC values for device No. 1 due to a technical problem occurring during the test day but not in the main SAPALDIA study. Further investigations revealed potential hardware and software sources of error which are not recognizable by trained technicians. These studies gave no evidence for systematic errors due to technician, team or device during the main SAPALDIA study. However, they revealed potential sources of error in modern devices, which function as "black boxes". Manufacturers should improve spirometry software to further enhance the technicians' attempts at accurate assessment.