The adsorption by Oxisols of 3 herbicides: atrazine (2-chloro-4-(ethylamino)-6-(isopropylamino)-s-triazine), terbutryn (2-(ter-butylamino)-4-(ethylamino)-6-(methyltio)-s-triazine), and 2,4-D (2,4-dichlorophenoxy acetic acid) was studied in two sites (MPR and MSP sites) with plots under natural vegetations and plots cropped for up to 56 years after clearing. The main mineralogical difference between the soils from the two sites was the higher content of gibbsite and hematite of the MPR than of the MSP site. In both locations, the soil organic matter (SOM) content decreased after clearing and cropping. The adsorption of atrazine and terbutryn (models of neutral and weakly basic herbicides. respectively) decreased with years of cultivation because of their significant correlation with decreasing SOM content. While the atrazine K(oc) values were relatively constant (average K(oc)=54+/-22 l.kg-1), the terbutryn K(oc) varied greatly with different soils (average K(oc)=308+/-184 l.kg-1). The terbutryn adsorption depended on soil pH: it was maximum between pH =4.5 and 5.0, and between 5.0 and 5.5 for soils from the MSP and MPR sites, respectively. The adsorption of 2,4-D (model of acidic herbicides) was always higher in soils from the MPR than from the MSP plots. In soils from the MPR site, 2,4-D adsorption decreased with years of cultivation. An opposite tendency was observed with soils from the MSP site. The adsorption of 2,4-D increased strongly with decreasing soil pH. Soils from the MPR plots were more sensitive to pH modifications than those from the MSP plots. In order to compare adsorption of polar herbicides on soils with different pH, we proposed to calculate the K(d) and K(oc) coefficients for a controlled pH.