Release of herbicides from lignin-hased formulations follows a diffusion-controlled mechanism. For mathematical modeling of diffusive transport, the conventional approach is to assume sink conditions at both surfaces of polymeric matrix. This boundary condition proved to be inadequate to describe experimental data obtained in a water dynamic bath system. However, satisfactory descriptions for this system were obtained when a stagnant unstirred layer of herbicide solution was used as the boundary condition. The adequacy of the model incorporating this new boundary condition was statistically tested using the Fisher test at a confidence level of 95% and plotting the residual distribution.