For decades X-ray diffraction (XRD) has been used as the standard method to determine the clay mineralogy of geological formations, including soils. For both pure and mixed-layer clay varieties, usual XRD identification methods rely on the position of diffraction maxima and on the shift of these positions as the result of the possible hydration/expansion properties of elementary layers. However, in low-temperature environments such as soils the frequent coexistence of different “phases” with similar compositions may reduce dramatically the efficiency of these techniques. Decomposition of XRD patterns allows overcoming this limitation by determining the position and the relative intensity of partially overlapped diffraction maxima. As a consequence, this method allows revealing subtle variations in the overall distribution of clay species (from the intensity ratio) and/or in their composition (from limited peak shift). However, a comprehensive determination of clay mineralogy must include a direct comparison between the experimental XRD data and a pattern calculated from the proposed identification so as to assess its validity (limits ?). Over the last few years, the trial-and-error optimization of the agreement between experimental and calculated XRD patterns (XRD profile modeling) has been increasingly used even for complex clay parageneses. This approach allows for a detailed structural characterization of both pure and mixed-layer clay phases and for a semi-quantitative phase analysis in complex mixtures. The two informations are essential to gain new insight into the actual nature of reactions taking place in geological environments. Additional constrains on the actual structure of the mixed-layer minerals present may be obtained by using several XRD patterns recorded on the same sample after different treatments (e.g. Ca-saturated in air-dried and/or ethylene-glycol solvated states). For a given sample, XRD patterns usually differ significantly after these treatments because of the contrasting hydration/expansion properties of expandable layers (multi-specimen technique).