Hydrocarbons and heavy metals are found in industrial polluted soils and are located in the solid phase of the soil. Due to different interactions with the matrix they can have an influence on the formation and the evolution of organo-mineral associations in the soil. The study is performed in order to test the hypothesis according to which hydrocarbons in soils control the aggregating processes and play an important role in specific sequestration of pollutants in stable aggregates. Moreover, plant roots promote fragmentation of aggregates, thus increasing the availability of pollutants and their uptake by the plants. Our aim is to study (1) the evolution in time of the aggregation in a polluted industrial soil, and (2) the role of vegetation on this process. The polluted soil was extracted from an iron and steel site, put in September 2005 in lysimeter of 3 m3 in an experimental station (GISFI) and studied according two treatments: (1) “bare soil”, for which a manual weeding is carried out and (2) “spontaneous vegetation” , for which no weeding was realized. The global amount of PAH is of 1551 mg/kg dry matter (total of 16 PAH). The approach consists (1) of an assessment of the structural stability, based on the use of soil fractionation techniques, allowing us to quantify the weights of the different organo-mineral fractions and to isolate them, and (2) a morphological and analytical characterization of these fractions at photonic and electronic scales. Results show that at the start of the experiment the sand fraction, which represents 70% of the soil, is associated with 10% of fine fraction (<50 µm) in water-stable aggregates. First macroscopic observations of different soil fractions reveal a great heterogeneity of size, shape and nature of the constituents. This is confirmed by the analyses at electronic scale. EDX analyses of the 0-2 µm fraction obtained after water-fractionation, showed smectites associated with Fe, Ca, K, but also Ca sulphates, organic matter and Fe oxides. Some particles were also identified as being essentially metallic and associated with organic matter. Some investigations are being done in order to evaluate the amount of PAH and metallic elements in the bulk soil and in the organo-mineral fractions. A second sampling will be effected in spring, and the same parameters will be studied. All these data enable us to specify the association of PAH and metals with the other constituents of such soils and bring useful information for a change of use of these sites.