Population studies in Britain and elsewhere report deficiencies in quality of pulmonary function measurements. Methods were tested to improve the standardisation of spirometry in an epidemiological study. The spirometer provided visual feedback about acceptability and reproducibility to American Thoracic Society (ATS) standards. After 14 weeks technicians (research nurses) were given feedback and further training. Measurements were repeated in a 5% sample. Participant characteristics and technical factors (technician and technician feedback) predicted unacceptable forced expiratory volume in 1 second (FEV1) and excessively variable FEV1 and forced vital capacity (FVC). Only participant characteristics predicted unacceptable FVC. Feedback to technicians reduced test failure for FEV1 by half and excessive within-session variability by one-third. In the reproducibility study, coefficients of variation for FEV1 and FVC were 3%. Epidemiological studies can achieve standards of between-session reproducibility for spirometry comparable to levels reported by pulmonary function laboratories. Performance feedback to technicians improves the level of minimally acceptable spirometry, and within-session reproducibility.