Sequential extraction procedures (SEPs) are widely used in environmental studies to infer the chemical and/or mineralogical forms of pollutants of concern in soils and sediments. Although there is no general agreement among the scientific community, these methods have shown some limitations, especially those with a lack of objectivity in their interpretation. In this work, a soil sampling campaign was carried out in an area affected by an abandoned Sb mine. Samples (0–15 cm) were carefully prepared and analyzed by an SEP. They were also studied by conventional mineralogical methods (optical and electronic microscopy, both scanning and transmission, with a coupled energy dispersive X-ray (EDX) spectrometer). When comparing the results obtained from both techniques, some discrepancies are highlighted, with As, Cu, Pb, Sb and Zn as elements of concern. For Sb, Cu, Pb and As the predominant fraction (excluding the residual one) is that associated with organic matter (from 8.54 for Sb to 18.90% for Cu). The fractions of pollutants linked to Fe and/or Mn oxides are quantitatively important for As, Pb and Zn (6.46%, 12.05% and 7.43%, respectively) and almost negligible for Cu and Sb. On the contrary, analyses carried out by EDX at a grain scale pointed out that no detectable quantities of the elements of concern were present on the surface of the organic particles. Sb and Pb were always detectable in Fe oxides (up to 1.84 and 5.76%, respectively). Regarding the role of the clayey fraction, the only disagreement between the employed SEP and the microanalysis is in relation to As. Arsenic bound to clay minerals was found to be an order of magnitude lower than As associated with Fe oxides (0.56% and 6.46%, respectively); in contrast, EDX microanalyses showed similar As contents in both groups. Given the objectiveness of EDX microanalysis, these differences should be considered inaccuracies in the interpretation of the sequential extraction results.