During FIFE 1987, surface energy fluxes were measured at 22 flux sites by nine groups of scientists using different measuring systems. A rover Bowen ratio station was taken to 20 of the flux stations to serve as a reference for estimating the instrument-related differences. The rover system was installed within a few meters from the host instrument of a site. Using linear regression analysis, net radiation, Bowen ratio, and latent heat fluxes were compared between the rover measurements and the host measurements. The average differences in net radiation, Bowen ratio, and latent heat flux from different types of instruments can be up to 10, 30, and 20%, respectively. The Didcot net radiometer gave higher net radiation while the Swissteco type showed lower values, as compared to the corrected radiation energy balance system (REBS) model. The four-way components method and the Thornthwaite type give similar values to the REBS. The surface energy radiation balance systems type Bowen ratio systems exhibit slightly lower Bowen ratios and thus higher latent heat fluxes, compared to the arid zone evapotranspiration systems. Eddy correlation systems showed slightly lower latent heat flux in comparison to the Bowen ratio systems. It is recommended that users of the flux data take these differences into account. ¿ American Geophysical Union 1992