The experimental reflectance data of Hueteet al. (1985) for field-grown cotton, beneath whose canopy soils differing widely in brightness had been inserted, were applied to two composite-scene reflectance models. In the additive-independent reflectance model (AIRM), the Sunlit soil, shadowed soil and Sunlit vegetation responses are considered as independent signals that are added linearly in porportion to the fractional ground area each component occupies. The transmittance-reflectance interactive model (TRIM) developed herein considers the reflectance of transmitted light up through the canopy from the soil as well as that reflected from the canopy itself. The TRIM model simulated the experimental measurements satisfactorily over the 0–100 per cent plant cover range for all soil backgrounds, whereas the AIRM model did not. The greenness lines closely paralleled the soil line for the AIRM model and thereby deviate from the measured lines at all plant covers. The greenness line slopes increased as the plant cover increased in the TRIM in agreement with experimental observations. The larger the proportion of shadows in the composite scene, the more important the reflectance-transmittance terms were in the TRIM. These results support other evidence that plant, soil and shadow reflectance components mix interactively to produce composite canopy reflectance.