The thermal conductivity of cultivation substrates has been one of the most important parameters in the heat transmission of ground under the thermal environment in a solar greenhouse. Among them, the cultivation substrate is one type of multicomponent material with solid and liquid phases. In this study, prediction models were proposed for the effective thermal conductivity of unsaturated cultivation substrates in the production of a solar greenhouse. Eight commonly-used single substrates were taken as the research objects, such as perlite, vermiculite, cinder, river sand, coco coir, peat, decomposed cow dung, and peanut shell. The thermal conductivities of the substrates were measured using the guarded hot plate apparatus. The test temperature was set at 20 ℃, where the temperatures of the hot and cold plates were 25 and 15 ℃, respectively. Firstly, the effective thermal conductivity of the solid phase in the eight single substrates in dry and saturation was determined to reversely calculate the hybrid models for the effective characteristics of composite materials, such as Series, Parallel, Johansen Geometric Mean, Maxwell-Eucken 1, Maxwell-Eucken 2, Effective Medium Theory, Series Parallel-Arithmetic Mean, Series Parallel-Harmonic Mean, Series Parallel-Geometric Mean, Maxwell Eucken-Arithmetic Mean, Maxwell Eucken-Harmonic Mean, and Maxwell Eucken-Geometric Mean model. The solid-phase thermal conductivities of perlite, vermiculite, cinder, river sand, coco coir, peat, and decomposed cow dung and peanut shell were 0.058, 0.139, 0.252, 0.817, 0.148, 0.518, 0.262, and 0.066 W/ (m·K), respectively. Secondly, the components of the solid phase were arranged in parallel using the forward and reverse calculation of the hybrid model for the effective characteristics of the composite materials, according to the measured effective thermal conductivity of the complex substrates. The least-square method was selected to verify the thermal conductivity of the six single models. It was found that the Parallel model presented the smallest among the six single models, such as Series, Parallel, Johansen Geometric Mean, Maxwell-Eucken 1, Maxwell-Eucken 2, and Effective Medium Theory model. Finally, the correlation analysis was made between the thermal conductivity of the solid phase and the volume proportion of each component in the composite substrates. Furthermore, a comparison was made on the experimental and theoretical values for the effective thermal conductivity of the composite substrates under different saturation degrees. The results show that the calculated values of the Parallel model were the closest to the measured ones, and the least-squares calculation presented the smallest among the six prediction models of effective thermal conductivity. Therefore, the Parallel model was suitable for the theoretical calculation and prediction for the effective thermal conductivity of the composite substrates. Moreover, four seedling and cultivation substrates were selected to verify the model, such as vermiculite:peat=8:2, cow dung:perlite:coco coir=1:1:1, peanut shell:vermiculite:perlite=5:3:2, and river sand:peat:vermiculite=2:1:3, commonly-used in practice under different saturation degrees. The results show that the mean absolute percentage error of the Parallel model was 0.026 2%～0.137 4%, the root mean square error of the parallel model was 0.008 6～0.031 5 W/(m·K), the model determination coefficient R2 of the Parallel model was 0.900 7～0.988 0, and the prediction accuracy of the Parallel model was high. indicating better performance. Consequently, the Parallel model can be used to accurately predict the effective thermal conductivity of the cultivation substrates in actual production under various saturation degrees in a solar greenhouse. This finding can provide a strong reference for the effective thermal conductivity of porous media in unsaturated cultivation substrates of a solar greenhouse. [ABSTRACT FROM AUTHOR]