OBJECTIVES: To assess the magnitude of error in pulmonary function measurements introduced by variation in spirometer temperature under field conditions. In a large scale epidemiological study of school children, the influence was investigated of spirometer temperature on forced expiratory volume in 1 second (FEV1) measured with dry rolling seal volumetric spirometers and conventional body temperature, pressure, and saturation (BTPS) corrections. METHODS: Linear regression analyses were performed on data from 995 test-retest pairs on 851 different children, with 1-110 days between test and retest, and spirometer temperature differences between -13 degrees C and +9 degrees C. RESULTS: After adjusting for effects of growth (test-retest intervals) and circadian variation (changes in times of testing), differences in standard BTPS corrected FEV1 showed significant (p < 0.05) dependence on differences in spirometer temperature between tests (-0.24%/degree C). CONCLUSIONS: When spirometer temperatures vary widely, standard BTPS correction does not fully adjust for gas contraction. To improve accuracy of volume measurements in epidemiological studies, additional correction for variation in spirometer temperature should be considered.