An intercomparison was made of the infrared gas analyzers used in the First ISLSCP Field Experiment (FIFE) 1989 to measure surface fluxes of carbon dioxide immediately following intensive field campaign (IFC) 5. Fluxes of CO2 agreed to within 15% when all instruments are compared; when models of the same type of analyzer were compared, the agreement was within 5%. The stability of the calibration coefficients is examined, and it is shown that frequent calibration of the gas analyzers is required. Differences in the physical design of the analyzer models discussed here, such as sampling volume and sensitivity of the electronics to thermal drift, influence the precision of the flux estimates. Spectra and cospectra reveal the degree to which instrumental noise limits the complete sampling of all frequencies of eddies contributing to flux transport. ¿ American Geophysical Union 1992