The variability of gas diffusion in soil is not well known, but is important for assessing greenhouse gas emissions, soil decontamination, oxidation in soil and plant and root respiration. The goal of this study was to assess small-scale variability of the relative soil-gas diffusivity (D s / D o , m soil air m −1 soil ) using large intact soil monoliths and to compare D s / D o calculation methods. Neon (Ne) was maintained constant at the lower boundary of three monoliths of two soils (a sand and an organic soil). Ne concentration was measured at large spatial and temporal frequencies. Calculation methods included the use of average concentration, and average D s / D o per horizon, per section, or for the entire soil profile. Considering all sections of the monoliths, D s / D o varied from 3.5 × 10 −3 to 1.2 × 10 −1 for the A p horizon and from 4.8 × 10 −3 to 8.3 × 10 −1 for the B f horizon in the sand and from 1.0 × 10 −3 to 7.9 × 10 −3 for the O hp horizon and from 2.4 × 10 −4 to 7.7 × 10 −2 for the O f horizon in the organic soil. For the entire soil profile, variations in D s / D o between monoliths reached 125% in the sand and 56% in the organic soil. The D s / D o calculation method influenced the apparent variability (CV) of D s / D o and, to a lesser extent, D s / D o values of the overall soil profile. Differences in D s / D o between monoliths could not be explained solely by the variability of total soil porosity and air-filled porosity. Soil macroporosity (cracks and earthworm burrows) and layering greatly influenced variability of gas movement. Thus, the choice of sampling procedure, calculation method and modelling must be governed by the scale of the processes of interest and soil variability attributes.