Several spirometer manufacturers use non-heated ceramic flow sensors to determine flow and calculate volume. Although some cooling of the air does occur as the air passes through the sensor, cooling is usually not complete. Therefore the usual technique is to apply a factor approximately equal to 30 percent of the full BTPS correction factor (CF). To determine the effectiveness of this approach, we tested several ceramic flow sensors with a mechanical pump using both room air and air heated to 37 degrees C and saturated with water vapor. The volume signals used to test the sensors were volume ramps and the first four ATS standard waveforms. Estimated BTPS CF for FVC and FEV1 were calculated by dividing the volume measured with room air by the volume measured with heated and humidified air. Our results using room air showed a considerable variability in the linearity of these flow sensors with one sensor showing a 400 ml difference (6.7%) in a 6 L volume ramp and flow rates of between 0.6 and 8 L/s. Using heated and humidified air, the estimated BTPS CF with the sensor initially at 20 degrees C ranged from 1.06 to 1.00 compared to a calculated value of 1.102. The estimated BTPS CF also varied with the number of curves previously performed, the time between curves, the volume of the current and previous curves, and the temperature of the sensor. Monitoring of the temperature of the air as it left the sensor (exit temperature) showed a steady rise in temperature with each successive curve. However, both the exit temperature and the estimated BTPS CF stabilized after approximately 5 curves using waveform 1 (FVC = 6L), provided there was only a short pause between curves. Use of exit air temperature alone proved to provide an effective means of estimating a dynamic BTPS CF. The use of a linear model - based on exit temperature - to estimate a dynamic BTPS CF, reduced the error in FEV1 to less than +/- 3 percent for exit temperatures from 5 to 28 degrees C. These results suggest that both linearization and dynamic BTPS CFs are needed for this type of flow sensor to operate within the ATS accuracy recommendations of +/- 3 percent for FVC and FEV1, particularly at lower operating temperatures.