Coal pyrolysis is highly dependent on the pore diffusion. However, classic diffusion laws, such as Fick's law and Knudsen's law, cannot accurately predict multicomponent gas diffusion during coal pyrolysis. To study multicomponent gas diffusion during coal pyrolysis, fractal pore models generated by a random walk algorithm were used to simulate coal particles, and the governing equations of multicomponent gas diffusion were established to model gas diffusion in fractal pore models. Many simulations were used to develop a diffusivity correlation for the fractal pores. The developed diffusivity correlation for fractal pores can be applied to multicomponent gas diffusion during coal pyrolysis. The diffusivity increases with increasing porosity, decreasing fractal dimension, and increasing equivalent pore diameter.