In respiratory health surveys involving multiple spirometers, spirometer differences may introduce important biases. We investigated temperature measurement variability as a cause of spirometer differences. Digital thermometers recorded internal (cylinder) and external (outer casing) temperatures of six similar rolling-seal spirometers during field use and in laboratory tests at controlled room temperatures. Internal and external thermometers substantially agreed in recording spirometer temperature changes, which lagged room temperature changes. Offsets of individual thermometers from overall mean readings were roughly the same in field testing of 3908 students in > 60 schools over 5 months and in subsequent laboratory tests. Thermometers differed by as much as 1.3 degrees C, causing differences as large as 0.8% in vital capacity measurements. We conclude that (1) interior and exterior temperatures of typical rolling-seal spirometers do not differ greatly, although both may differ from surrounding air temperature; and (2) variations between individual digital thermometers may be large enough to bias spirometric data appreciably in large-scale surveys. Variations should be controlled by selection of similar-reading thermometers and/or correction to a uniform standard.