We propose a novel methodology that represents die’s local metal density-dependent anisotropic thermal properties. Each layer is considered the binary composite that consists of metal lines embedded in an insulator. The die is divided into smaller bins in which local metal densities are calculated. Each bin is represented by own thermal properties that depend on local density and metal routing direction. These bins act as building blocks for constructing die’s thermal netlist with improved accuracy. The generated thermal netlist has been employed for die thermal analysis, and has demonstrated excellent agreement with finite element method, with much faster runtime.