Atmospheric and land surface data collected from the HAPEX-MOBILHY field experiment were used to compare the bare soil evaporation simulations of 13 land surface schemes and to examine the relationship between differences in evaporation and differences in soil moisture. For a 120-day period in which there was no vegetation present, computed total evaporation ranged between 100 and 250 mm. This large range in evaporation was not related to soil moisture differences. Prescribing surface soil moisture and temperature did not reduce the range in evaporation and instead the range was increased. The models' predictions of evaporation were in closer agreement with each other when they were allowed to select their own surface conditions than when they were forced to use the same conditions. The bare soil evaporation formulations used by the land-surface schemes are not consistent with each other and these inconsistencies produce widely-varying bare soil evaporation rates. The range in bare soil evaporation is unlikely to be reduced by improving the simulation of soil moisture and instead an assessment of why the bare soil evaporation formulations are inconsistent is required.