Comprehensive Monte Carlo simulations are used to correct deviations in the atmospheric downwelling longwave (DLW) radiative flux calculated by isotropic scattering assumptions. The widely used $δ$-M approximation is validated for low to medium values of aerosol loading. For very high aerosol loading conditions, the $δ$-M approximation incurs an error. Here we propose scaling corrections for extreme loading conditions routinely found in selected urban areas in Asia, but also in other continental and coastal areas susceptible to large-scale wildfire pollution (Western USA) or dust storms (Mediterranean regions and Northern Africa). The scaling rules are expressed as functions of the normalized aerosol optical depth t ∗ and the scattering asymmetry factor e g . An exponential relationship between the DLW deviation that assumes isotropic scattering and t ∗ is found, and the corresponding fitting coefficients are correlated for different types of aerosols (sample internal mixing, urban, continental and marine aerosols). The $δ$-M approximation is sufficiently accurate when aerosol optical depths (AOD) at the ground level are smaller than 0.5. For AOD beyond this threshold, the proposed scaling rule corrections should be used for estimation of downwelling thermal radiative fluxes. The effects of moisture content on aerosol composition and on DLW radiative fluxes are also investigated for all conditions of interest.