This study proposes a model for self-diffusion coefficients of pure substances from entropy scaling using the perturbed-chain polar statistical associating fluid theory (PCP-SAFT) equation of state. In accordance with the entropy scaling approach proposed by Y. Rosenfeld [Rosenfeld, Y. Phys. Rev. A 1977, 15, 2545-2549], we observe that the self-diffusion coefficient of real substances, once made dimensionless with an appropriate expression, only depends on residual entropy. The proposed model requires 3 parameters for each pure substance. For substances with scarce experimental data, however, a scheme is proposed to estimate one or two of these parameters. We study 133 substances from more than 14 different chemical families and find the average absolute deviation of 8.2% between the proposed model and experimental data (9992 data points). The model shows satisfying robustness for extrapolating self-diffusion coefficients to conditions rather distant from the state points where experimental data are available.