Competitive sorption of protons, Cu, and Pb onto kaolinite (KGa-2) was investigated over wide concentration ranges and quantitatively described using three different models based on surface complexation and cation exchange reactions. In all models, two types of binding sites were assumed for kaolinite: edge sites (SOH0.5-) with pH-dependent charge and face sites (X-) with permanent negative charge. In a first step, proton sorption was measured by potentiometric acid-base titrations of kaolinite dispersed in 0.01, 0.03, and 0.1 M NaNO3 electrolyte solutions. The acid-base titration data were fitted to obtain site densities and protonation constants for the edge and face sites, respectively. In a second step, the sorption of Cu and Pb onto kaolinite was investigated at fixed pH values by metal titration using ion-selective electrodes for Cu2+ and Pb2+, respectively, and by independent batch sorption experiments. Our metal sorption data cover a range of pH 4-8 for Cu and pH 4-6 for Pb, three different ionic strengths (0.01, 0.03, and 0.1 M NaNO3), and up to eight orders of magnitude in free metal ion activity. An additional experiment was conducted to explore the sorption competition between Cu and Pb. In all three models, sorption of protons and metal cations to the edge sites of kaolinite was described with a 1-pK basic Stern (BS) approach. The three models differed only in the description of cation sorption to the face sites. In the first model (BS/GT), we used a Gaines-Thomas (GT) cation exchange equation for the face sites. This model yielded a satisfactory description of Cu sorption, but failed to describe Pb sorption isotherms at pH 4, 5, and 6. In the second model (BS/BS), we replaced the Gaines-Thomas equation by a basic stern surface complexation formulation. thereby introducing electrostatic terms for sorption to face sites and allowing for free binding sites X-. This did not improve the fits of Cu or Pb sorption to kaolinite, however. In the third model (BS/BSext) we extended the BS/BS-model by introducing additional monodentate sorption complexes at face sites (X-Cu+ and X-PbNO3). This model described both Cu and Pb sorption very well over the entire range in metal concentrations and pH. It also correctly predicted the competitive effect of Pb on sorption of Cu. Model calculations with all three models suggested that Cu and Pb were sorbed mainly to face sites at low pH, while sorption to edge sites dominated at high pH values. (C) 2004 Elsevier Inc. All rights reserved.