By means of a fundamental solution for a single inhomogeneity embedded in a functionally graded material matrix, a self-consistent model is proposed to investigate the effective thermal conductivity distribution in a functionally graded particulate nanocomposite. The “Kapitza thermal resistance” along the interface between a particle and the matrix is simulated with a perfect interface but a lower thermal conductivity of the particle. The results indicate that the effective thermal conductivity distribution greatly depends on Kapitza thermal resistance, particle size, and degree of material gradient.