Soil desiccation (drying), induced by dry air injection and moist air extraction, is a potentially robust remediation process to slow migration of inorganic or radionuclide contaminants through the vadose zone. The application of gas-phase partitioning tracer tests has been proposed as a means to estimate initial water volumes and to monitor the progress of the desiccation process at pilot tests and field sites. In this study, tracer tests were conducted in porous medium columns with various water saturations using SF6 as the conservative tracer and trichlorofluoromethane and difluoromethane as the water-partitioning tracers. For porous media with minimal silt or organic matter fractions, tracer tests provided reasonable saturation estimates for saturations close to zero. For sediments with significant silt or organic matter fractions, however, tracer tests only provided satisfactory results when the water saturation was greater than 0.1 to 0.2. For drier conditions, the apparent tracer retardation increased due to air–soil sorption, which is not included in traditional retardation coefficients derived from advection–dispersion equations accounting only for air–water partitioning and water–soil sorption. Based on these results, gas-phase partitioning tracer tests may be used to determine initial water volumes in sediments, provided that the initial water saturations are sufficiently large. Tracer tests are not suitable for quantifying moisture content in relatively dry sediments, however, especially if significant amounts of organic matter or silt are present.