Abstract: Determination of soil surface properties relies on the study of adsorption isotherms, and most commonly nitrogen adsorption isotherms (NAI) are monitored for this purpose. In general, simply the so-called BET model is used to mathematically describe adsorption isotherms, which provide soil specific surface area (SSA) estimates. Useful data for computing SSA by the BET approach are restricted to a limited range of the experimental isotherm. On the other hand, the entire NAI curve has been analyzed by fractals and more recently also by multifractal models. The main objective of this work was to characterize NAI from tropical soils collected in Brazil using the multifractal approach. In addition, the variability of both SSA and multifractal parameters as a function of soil physical and chemical properties was accounted for by statistical and principal component analysis (PCA). Soil type, profile differentiation, and local geology were used to highlight the interpretation of PCA results. Nitrogen adsorption isotherms were determined in duplicate for 54 horizons collected from 19 soil profiles in Minas Gerais state, Brazil. Ten of the studied profiles were classified as Ferralsols (Latosols, Oxisols). Besides this main soil group, other widely different soil groups were sampled, including Nitisol, Acrisol, Alisol, Luvisol, Planosol, Cambisol, Andosol, and Leptosol. Both soil SSA and the cumulative N2 adsorbed by a relative pressure of 0.95 (V0.95) were positively and significantly correlated with clay content, Fe2O3 and Al2O3. The scaling properties of the NAI curves from all the studied soil horizons could be fitted reasonably well with multifractal models. Analysis of singularity spectra showed mean values of Hölder exponent of order zero, α0, ranging from 1.229 to 1.869. Mean values of the entropy dimension, D1, obtained from generalized dimension spectra varied from 0.317 to 0.749. These results indicate that the probability mass distribution function is highly heterogeneous, and various degrees of heterogeneity between NAI from the studied horizons occur. Principal component analysis and analysis of variance showed that multifractal parameters were appropriate to associate profiles, soil sub-groups, and groups with similar properties. Moreover, PCA results provided further evidence showing that SSA and NAI multifractal parameters clearly describe different properties of the soil surface.