This paper describes the course of topsoil thermal conductivity, λ, diffusivity, α, and heat capacity, C h , during two measurement campaigns, conducted in semi-arid areas--the EFEDA-I experiment and HAPEX-Sahel.For the derivation of α, five methods (the Amplitude, Phase, Arctangent, Logarithmic and Harmonic equation) were compared. Values of C h were derived from measurements of soil moisture content, θ, and dry bulk density. λ was either measured directly (the non-stationary probe method) or calculated from λ = α C h .Thermal soil properties were clearly related to θ (and thus rainfall) throughout the measurement campaigns: hardly any changes occurred during the EFEDA-I experiment where continuous dry conditions prevailed, whereas for HAPEX-Sahel a clear decrease in all thermal properties was observed after the last rainfall.For calculation of α, the Amplitude and the Harmonic equation gave the best results. Calculation of α and λ beneath the vegetation plots yielded unreliable results, mainly due to shading effects causing more than one temperature maximum.Direct measurement of λ yielded unrealistically low values for the dry soil conditions as encountered in Spain, due to poor contact between probe and soil. A correction for the contact resistance is necessary to obtain better estimates. For HAPEX-Sahel, measured and calculated λ values were much closer, mainly for reasons of higher θ values and a high soil compaction which ensured better contact between soil and probes.