We undertook an interlaboratory comparison of techniques used to extract and analyze trapped gases in ice cores. The intercomparison included analyses of standard reference gases and samples of ice from the Greenland Ice Sheet Project 2 (GISP2) site. Concentrations of CO2, CH4, the Δ18O of O2, the Δ15N of N2, and the O2/N2, and Ar/N2 ratios were measured in air standards and ice core samples. The standard reference scales for CO2 and CH4 were consistent at the ¿2% level. The ΔO2/N2 and Δ18O of O2 measurements showed substantial deviations between the two laboratories able to measure these ratios. The deviations are probably related to errors associated with calibration of the working standards. The ΔAr/N2 and Δ15N of N2 measurements were consistent. Five laboratories analyzed the CH4 concentration in a 4.2-m section of the GISP2 ice core. The average of 20 discrete CH4 measurements was 748¿10 parts per billion by volume (ppbv). The standard deviation of these measurements was close to the total analytical uncertainty associated with the measurements. In all cases, those laboratories employing a dry extraction technique determined higher CH4 values than laboratories using a wet extraction technique. The origin of this difference is unclear but may involve uncertainties associated with blank corrections. Analyses of the CO2 concentration of trapped gases showed extreme variations which cannot be explained by analytical uncertainties alone. Three laboratories measured the <CO2> on 21 discrete depths yielding an average value of 283¿13 parts per million by volume (ppmv). In this case, the standard deviation was roughly a factor of 2 greater than the analytical uncertainties. We believe the variability in the measured <CO2> results from impurities in the ice which may have compromised the <CO2> of trapped gases in Greenland ice. ¿ 1997 American Geophysical Union