Even though the X-ray diffraction (XRD) is a usual method for soil mineralogical analysis, there is still no consensus regarding the optimal sample preparation method: is it better to use combustion in order to remove the organic phase and extract the mineralogical phase, or is it preferred to perform the XRD analysis on a minimally prepared, almost pristine sample, despite the enormous complexity of soil as a matrix? In this work we thus analysed two coastal and one continental soil. For the first series of measurements, the samples were just sieved through the 1-mm mesh after removing the coarse soil constituents. For the second series, the samples were combusted for 8 h at 480 °C, and the resulting ash was analysed. The optimisation of sample preparation, and the subsequent XRD data analysis was performed in order to determine the qualitative soil mineralogical composition, as it represents one of the most influential feature determining soil physico-chemical properties. For coastal soils, quarts was established as a dominant phase, in accordance with a large sand percentage common in those soils, while different silicon dioxide polymorph dominates the continental soil composition. Clays were found in all the soils, but of different kinds: muscovite for a continental soil, montmorillonite and nacrite for coastal soils. The results also showed pronounced differences between pristine soils and ash samples after combustion. The combustion caused thermal degradation of some components, phase transitions to different polimorphs, and elevated amounts of amorphous phase. Ash difractograms contained phases undetected in pristine samples’ results. The results thus indicate that sample preparation is a crucial factor determining the relevancy of the XRD analysis, the best option being the minimal sample preparation and avoidance of combustion.