The performance of the Metamax I and the Metamax II portable analysers for measuring the O2 uptake has been examined during exercise. Healthy subjects ran on the treadmill or bicycled on ergometers while the O2 uptake was measured by the Metamaxes and also by the Douglas bag technique or the Vmax 29 instrument. In the first series of experiments, O2 uptake was measured by each instrument in turn. In later experiments two or more breathing valves were connected in a series, thus enabling measurement of the O2 uptake simultaneously by more than one instrument. The O2 uptake measured by the Metamax analysers rose linearly by the value given by the control methods. However, there were variations of approximately 5% because the relationships differed between subjects. When the data from each subject were examined separately, the error of regression was 0.5-1 micromol s(-1) kg(-1) (2-3%), and the error of regression when relating the O2 uptake to the exercise intensity was similar to that found when using the Douglas bag technique alone. In most cases the lung ventilation reported by the Metamaxes was a few percent less than that given by the control methods, while the fractional extraction of O2 was higher for the Metamaxes. The respiratory exchange ratios (R-value) reported by the Metamaxes were in good agreement with those of the control methods in the range 0.9-1.0 only; for this parameter, the Metamaxes do not seem to be reliable for exercise testing. The O2 uptake and the R-value were also calculated from the raw data reported by the Metamaxes. The calculated values differed somewhat from those reported by the instruments, and the calculated values were more in agreement with those obtained by the Douglas bag technique than those reported by the instrument. This study suggests that the O2 uptake reported by the Metamaxes is precisely measured within subjects but that there are some systematic errors as well as variations between subjects.