This paper describes an improved numerical method to estimate the gas diffusion coefficient (D) of a finite soil sample, giving as a by-product an estimation of the effective porosity (e). This method is applied to an earlier apparatus used to measure the rate of transfer of tracer krypton-85 through a finite soil sample: the soil sample is enclosed by two gas cells, and the concentration in each gas cell is regularly measured after gas injection using β radiation count rates from the test gas. The proposed numerical approach combines the finite-element method to solve Fick's second law of diffusion, and a nonlinear, iterative procedure to find the estimated parameters&OV0429;andê, where the residual differences between the measured and simulated count rates at opposite ends of the sample are minimized.An error analysis that takes into account the random process of β emission of tracer is numerically simulated: the gas diffusion coefficient is shown to have a low sensitivity to this source of error, whereas the affective porosity has a larger sensitivity.The proposed method is applied to a large range of soils, including (1) wet and dry undisturbed field soil cores, (2) compacted, water-saturated aggregates, (3) dry and wet textural soil cores. The parameters&OV0429;andêwere fitted satisfactorily to the measured data on each soil sample: the diffusion coefficient estimates are compared with existing estimation methods, whereas the estimated effective porosity is approximately the same when using the iterative proposed approach or a mass balance method applied to the experimental apparatus. Furthermore, it is shown that there is no simple and unique relationship for all the porous media between the calculated gas diffusion coefficient and soil sample air-filled porosity: such a relationship is likely to be highly dependent on the physical characteristics of the pore space, such as pore continuity, tortuosity, and morphology.