In order to study temporal and spatial variations of atmospheric CH 4 quantitatively, we originally improved a measurement system for carbon and hydrogen isotopic ratios (d 13 C and dD) of CH 4 to attain high-precision measurements. By analyzing 100 mL aliquots of an ambient air sample, the precision of our system is 0.080 for d 13 C and 2.20 for dD (1s), which are one of the highest precisions reported so far. The system consists mainly of a CH 4 preconcentration device and a continuous-flow gas chromatograph isotope ratio mass spectrometer equipped with a combustion furnace and a pyrolysis furnace for measurements of d 13 C and dD. The preconcentration trap temperature was maintained at À130 G 1 C during collection of CH 4 from the air sample by passing it through the trap, then at À83 G 1 C while remaining air components such as N 2 and O 2 except for CH 4 escaped, and finally at 100 G 1 C for CH 4 elusion. The isotopic values are measured on a mass spectrometer, relative to respective reference gases. For this study, the d 13 C and dD values of the reference gases were calibrated against our primary standards provided by the IAEA: our d 13 C primary standard is NBS18, whereas our dD primary standards are V-SMOW and SLAP. To ensure the long-term stability and reproducibility of our measurement system, a calibrated whole air stored in a high-pressure cylinder, which was called ''test gas,'' was measured at least twice on each day when sample measurements were made. To measure small air samples, such as those extracted from ice cores, we also examined the relation between the sample size and the measured value of d 13 C and dD: gradual enrichment of the d 13 C occurred with decreasing CH 4 content less than 8 nmol whereas no such e¤ect could be seen for the dD. Furthermore, preliminary results of latitudinal distributions of d 13 C and dD were discussed along with CH 4 concentrations obtained by our shipboard air-sampling program.