Soil thermal conductivity (λ) is an important thermal property for environmental, agricultural, and engineering heat transfer applications. Existing λ models for frozen soils are complicated to use because they require estimates of both liquid water content and ice content. This study introduces a new approach to estimate λ of partially frozen soils from air-filled porosity (n a), which can be determined by using an oven-drying method. A λ and n a relationship was established based on measurements for 28 partially frozen soils. A strong exponential relationship between λ and n a was found (with R 2 of 0.82). Independent tests on 10 partially frozen soils showed that the exponential λ-n a model produced reliable λ estimates with a RMSE of 0.319 W m −1 K −1 , which was smaller than those of two widely used λ models for partially frozen soils. The λ-n a model is easier to use than existing models, because it requires fewer parameters. Note that the λ-n a model ignores the effect of temperature on λ of frozen soils and is most applicable to soil at temperatures of at least −4 • C. 1 INTRODUCTION Soil thermal conductivity (λ) is a key parameter in modeling heat transfer in the vadose zone, which is impor-Abbreviations: RMSE, root mean square error; SE, standard error of the regression; TDR, time domain reflectometry.