Between five and eight metres of a soil profile (Oxisol) were removed during the building of a power station in Ilha Solteira-SP (Brazil). This subsoil was tested for four years with different revegetation treatments, in some cases emendations were applied. Treatments: T1: (control, exposed soil), T2: uncultivated soil, T3: Mucuna (Mucuna aterrima), T4: guandu (Cajanus cajan)/beans, T5: limed+mucuna, T6: limed+guandu, T7: limed+gypsum+mucuna, T8: limed+ gypsum + guandu. Afterwards Braquiaria decubems was sowed. Six years later, three depths were sampled (0-10, 10-20 y 20-40 cm) in each testing part (four repetitions). The aggregates stability was evaluated (wet sieving) and the aggregate mean weight diameter (MWD) was determined. Between 0-10 cm of depth, the MWD of the aggregates stable to water was higher than in the other depths (10-20 y 20-40 cm), partly influenced by the effect of the brachiaria's radicled system, concentrated mainly in this layer. The following sequence was observed in relation to the treatments: T8 > T6 > T7 > T3 > T2 > T4 > T5 > T1. The control plot (T1, exposed soil without vegetation), showed the lowest MWD value (1.76 mm) proving more inestability than the other plots in recuperation process. However, the treatment T8 appeared to be the most effective to recover the aggregates stability, that show the MWD value (2.64 mm). Likewise, T6 and T7 treatments showed themselves MWD similar values, 2.46 y 2.40 mm respectively.The behaviour of aggregates in plot T2, the one that was tilled but not cultivated, was similar to T3 (mucuna before braquiaria). This could have to do with the copious spontaneous vegetation developed throughout several years that would provide organic matter, and together with the effect of this type of roots would help in the aggregation process. Despite that the treatments to recover this degradated area allowed to recover most of the aggregates stability, it is less than the normal levels found in nearly cultivated areas.