Mass spectrometers are used in ICUs and ORs to measure the concentration of medical and anesthetic gases gathered from multiple sites. This investigation was designed to determine the accuracy of a clinical system, which included 12 ICU bedside stations monitored by a medical mass spectrometer (Perkin-Elmer RMS III, Pomona, CA). Each site station was connected to the analyzing unitviatwo Teflon tubes, one permanently installed, 30-m long, and the second disposable, 2.4-m long. A gas mixture containing 95% O2and 5% CO2, alternating with room air, was delivered to a solenoid valve and from there to the connecting tubes. Gas flow-rate, delay time, rise time, and peak and through concentrations were determined for each gas at solenoid cycling frequencies of 25, 50, and 100/min. After the first set of measurements, the 30-m tubes were thoroughly cleaned and all measurements repeated. In addition, the authors also measured CO2delay and rise times when the gas was delivered to the mass spectrometer through an unused 30-m tube or a new 2.4-m tube.Gas flow-rate increased from 143 ± 1 ml/min (mean ± SD) to 238 ± 9 ml/min after the tubes were cleaned. Delay time was identical for all gases at all solenoid cycling rates but decreased significantly (P< 0.05), from 11.5 ± 0.3 to 4.8 ± 0.7 s after the ceiling tubes were cleaned. As solenoid valve rate increased, the difference between measured and actual gas concentration increased. The lowest accuracy was 63.6 ± 2.1%, for CO2at 100 cycles/min. Even after the tubes were cleaned, peak measured values of CO2concentration were only 82.9 ± 4.7% of steady-state concentration at the highest solenoid valve cycling rate. When CO2passed through the tubes connecting active beside stations to the mass spectrometer, rise time was 583 ± 14 ms. This value decreased to 453 ± 84 ms after the ceiling tubes were cleaned but remained significantly higher than when new 30-m (272 ± 3 ms) or 2.4-m (178 ± 2 ms) tubes were used.The authors conclude that gas concentration measured by mass spectrometry can be quite inaccurate when gases are sampled from remote stations, especially at respiratory rates ≥ 50/min. This problem is largely related to the obstruction caused by the progressive accumulation of particulate materials in the long gas transport tubes. These errors cannot be detected by static self-calibration routines, but require dynamic testing.