A commercially available catheter type electrode with which PCO2 can be continuously measured in vivo and in vitro gave progressively less accurate results the longer the measuring period was extended. This proved to be due to temperature effects and a change in sensitivity with time. A correction procedure for these effects was developed which was based on two observations. 1. The relationship between temperature and the logarithm of the sensitivity of the electrode-amplifier combination was linear and virtually identical for 9 electrodes: 8% change in sensitivity for a deviation of 1 degree C from the temperature during calibration. 2. The change in sensitivity due to drift of the electrode output is approximately a logarithmic function of time: 1 h after calibration all electrodes exhibited a decreased sensitivity, varying between 0.3 and 16.7%. The drift effect can be dealt with by repeated calibrations, preferably at 1 1/2 h intervals. The adequacy of the correction procedure was assessed in in vivo measurements in cats and dogs. The mean PCO2 difference between the in vivo measurement, corrected for temperature and drift, and samples analyzed with a conventional electrode, was 0.005 kPa (0.04 mm Hg) with a standard deviation of 0.187 kPa (1.39 mm Hg).