The purpose of this work is to test a mechanistic model for predicting depth profiles of stable isotope values in unsaturated soil under natural climatic conditions. An experiment was conducted on a (20¿50 m) bare field (27% clay, 62% silt, and 11% sand) at Montfavet (France) for a 50-day period. The field was irrigated with a sprinkler (approximately 7 mm h-1) at the beginning of the experiment (July 9), after which evaporation operated under natural conditions (July 10). Three to five gravimetric water content profiles, down to 0.20 m, were measured daily. The field was equipped with tensiometers at 0.10-m intervals from 0.15 to 1.55 m depth at three sites. Temperature probes were installed at intervals down to 1.00 m depth at one site, and soil-temperature measurements were recorded every 15 min. Soil-water was sampled for isotopic analysis about every 5 days. A micrometeorological mast was set up in the center of the field to measure wind speed, air temperature, and vapor pressure and to estimate the sensible heat flux (eddy correlation method). Unsaturated hydraulic conductivity and aerodynamic roughness length were estimated before running a full comparison with field data. Good agreement was achieved between observed and calculated hourly fluxes for net radiation, soil heat conduction, sensible heat flux, and actual soil evaporation. The model described well the variations of both moisture content and soil temperature. The isotopic enrichment in the top soil layers was qualitatively well described, although systematic differences appeared in the shape of the calculated and measured isotope profiles. The model results were highly sensitive to small changes in convective transport in the liquid phase and to the initial isotope profile. More intensive soil sampling would be useful for validating the model. ¿ American Geophysical Union 1996