The purpose of this study was to investigate the characteristics of a newly developed turbine flowmeter (Alpha Technologies, model VMM-2) for use in an exercise testing system by comparing its measurement of expiratory flow (VE), O2 uptake (VO2), and CO2 output (VCO2) with the Fleisch pneumotachometer. An IBM PC/AT-based breath-by-breath system was developed, with turbine flowmeter and dual-Fleisch pneumotachometers connected in series. A normal subject was tested twice at rest, 100-W, and 175-W of exercise. Expired gas of 24-32 breaths was collected in a Douglas bag. VE was within 4% accuracy for both flowmeter systems. The Fleisch pneumotachometer system had 5% accuracy for VO2 and VCO2 at rest and exercise. The turbine flowmeter system had up to 20% error for VO2 and VCO2 at rest. Errors decreased as work load increased. Visual observations of the flow curves revealed the turbine signal always lagged the Fleisch signal at the beginning of inspiration or expiration. At the end of inspiration or expiration, the turbine signal continued after the Fleisch signal had returned to zero. The "lag-before-start" and "spin-after-stop" effects of the turbine flowmeter resulted in larger than acceptable error for the VO2 and VCO2 measurements at low flow rates.