Soil moisture is a very important component of the climate system, regulating both energy and water fluxes on local and regional scales. To obtain accurate climatological and meteorological forecasts we therefore need to obtain accurate predictions of soil moisture from land surface models. Actual in situ measurements are crucial for testing and developing these models. It is to this end that we have obtained a new soil moisture dataset from the Boissy-le-Châtel hydrological station in the Orgeval catchment in France. This dataset has a time span of over 5 years, from late 1996 to early 2003. It includes soil moisture measurements for 11 layers, as well as meteorological observations which can be used to force a land surface model. The data will be made available to the public via the Rutgers University Center for Environmental Prediction Global Soil Moisture Data Bank. We force the NOAH land surface model with the meteorology observations, and evaluate the resulting soil moisture, sensible and latent heat fluxes, and runoff using the Boissy-le-Châtel data. This experiment can serve as a prototype for other models to conduct their own evaluations, in the spirit of PILPS Phase 2.