As erosion, exacerbated by the extension of vine growing on hillslopes in Mediterranean environment, takes part in the spatial reorganization of topsoil and his carbon stock, indicators stemmed from this phenomena could be useful for studying space-time variations of carbon stocks. On the other hand, aggregate stability indexes could be useful also to define such indicators because they can be relevant in this environment inducing runoff and erosion when soils are uncovered or low plant covered. Within this framework, our objective was to test four widely used methods for aggregate stability determination, that includes slaking effects and gives indexes such as macro-aggregates rates (MA 200) or mean weight diameter (MWD). The test was performed in three French Mediterranean study sites characterized by brown calcareous soils, vineyard and the existence of previous erosion data from simulated rainfalls.Our first goal was to select the aggregate stability method which was best correlated with datas from simulated rainfalls. For that, we chose 5 farming situations, corresponding to 14 simulated rainfalls with an intensity of 60 mm/h, located on the stations of Corconne village and Pradel. Significant relationship were found between rainfall simulations data and aggregate stability indexes : the relationships are positive and linear in case of the runoff initiation variable and negative and curvilinear in case of the other runoff and soil losses variables. The 2 indexes resulting from the method of Le Bissonnais (MWD in its logarithmic form and MA 200) are the best adjusted with all these variables. The second goal was to analyze the relationship between aggregate stability and soil properties that were relatively easy to spatialize. We took complementary soil samples on various situations of the 3 selected sites ; thus, 68 soil samples were carried out on the whole. The relationships between the 2 indexes of Le Bissonnais's method and soil properties show that there is a very significant correlation with the organic carbon rate (CSOM), while the adjustments are significantly better with curvilinear regression models. The 2 indexes of Le Bissonnais's method, that give very significant correlations with simulated erosion data and CSOM which is a good indicator for the spatialization of these 2 indexes, appears to be very useful for the forecasting of the spatial evolution of carbon stocks in the studied area.