Soil thermal conductivity (λ) and heat capacity (C) control heat transport and the thermal environment for biogeophysical processes in the vadose zone. Accurate λ and C predictions for peaty soils with high organic contents are particularly important for assessing emissions of greenhouse gases formed during microbial activity in wetlands. In this study, we measured the λ and C at different soil-water matric potentials on undisturbed samples for three peaty soil profiles at the Hokkaido Bibai marsh in Japan, representing a total of 10 different soil horizons. The thermal properties under air-dried conditions, λdry and Cdry, were measured separately at changing volumetric solids contents (σ). For each sample, volume shrinkage was observed to varying degrees during the drying process. Measured λ and C increased linearly with increasing volumetric water content (θ). Applying the concept of a three-phase mixing model and incorporating the λ–θ or C–θ and the λdry–σ or Cdry–σ relations, predictive λ and C models were developed as functions of σ and θ . The new mixing model is represented by λ = λdry + fλθλw and C = Cdry + fCθCw, where λw and Cw are the thermal conductivity and heat capacity of water, respectively, and f is an impedance factor that takes into account the liquid-phase tortuosity. The new mixing model predicted literature λ–θ data on peaty and highly organic soils under variable saturation well. The probable ranges of λ and C under variable saturation were proposed based on the sensitivity analysis.