Transport modelling, risk assessment, and the evaluation of remediation strategies at sites contaminated with volatile organic compounds (VOCs) require the knowledge of gas-phase diffusion coefficients in unsaturated porous media. This thesis presents a new in situ method for the simultaneous measurement of effective and sorption-affected diffusion coefficients. The method relies on the injection of a gaseous compound and a conservative gaseous tracer using a soil gas probe. Concentrations are monitored at the tip of the probe during 8 hours and evaluated with an analytical equation for reactive transport in a homogeneous medium. Results are reported for volatile organic pollutants in a sand-filled lysimeter and at a field site with a sandy unsaturated zone. The determination of the effective diffusion coefficients showed good reproducibility and robustness with respect to analytical error, with an estimated overall error of 13%. The determination of the sorption-affected diffusion coefficients also showed good reproducibility. However, batch experiments suggest a systematic overestimation by 15-20%, because the tracers were not truly conservative. The in situ method avoids considerable uncertainties associated with estimating apparent diffusion coefficients from empirical relationships and avoids problems with establishing representative soil conditions in laboratory experiments. Another key issue for risk assessment at contaminated sites is the localization and quantification of nonaqueous phase liquids (NAPLs) such as gasoline, kerosene or heating oil in the subsurface. This thesis proposes the theory and practical application of a new diffusive partitioning tracer test (DPTT) for NAPL detection in the vadose zone. The general approach is the same as for the determination of apparent diffusion coefficients. A mixture of chlorofluorocarbons as gaseous tracers is injected into the vadose zone using a soil gas probe. While the tracers diffuse away, small volumes of gas are withdrawn from the injection point. The quantitative determination of the NAPL saturation is based on a comparison of the concentration decline of tracers with different air-NAPL partitioning coefficients. An alternative approach is the measurement of diffusive breakthrough curves in some distance from the injection point. The test has been evaluated in laboratory sand columns, at a field site with an artificial contamination embedded in a sandy unsaturated zone and in a sand-filled lysimeter, where the NAPL distribution was heterogeneous. NAPLs in saturations between 0 and 4 % of the total porosity have been reliably detected and quantified in a wide range of different water contents. NAPL saturations determined by DPTTs and by extraction from soil cores agree within a factor 2 or better. Results from the lysimeter study suggest that the NAPL saturation derived from the tracer concentrations at the injection point represent a small soil volume in the vicinity of the injection point, whereas the NAPL saturation determined from diffusive tracer breakthrough curves represent the soil between injection and monitoring point. These innovative investigation methods for the vadose zone are inexpensive, require little technical equipment on site, and allow the simultaneous determination of different parameters at one location from a few single measurements: as demonstrated in situ for two sandy soils, the effective and the apparent diffusion coefficient or the NAPL saturation, the effective diffusion coefficient and the soil gas concentration of several VOCs can be determined simultaneously by the proposed methods. This new approach has been tested in sandy, unsaturated porous media. The performance of the methods in heterogeneous environments or in porous media other than sand needs to be further investigated.