The adsorption of cobalt on samples from a potential waste repository site in an arid region was investigated in batch experiments, as a function of various solution phase parameters including the pH and ionic strength. The samples were characterized using a range of techniques, including BET surface area measurements, total clay content and quantitative X-ray diffraction. The statistical relationships between the measured cobalt distribution coefficients (K(d) values) and the solid and liquid phase characteristics were assessed. The sorption of cobalt increased with the pH of the aqueous phase. In experiments with a fixed pH value, the measured K(d) values were strongly correlated to the BET surface area, but not to the amount of individual clay minerals (illite, kaolinite or smectite). A further set of sorption experiments was undertaken with two samples of distinctive mineralogy and surface area, and consequently different sorption properties. A simple surface complexation model (SCM) that conceptualized the surface sites as having equivalent sorption properties to amorphous Fe-oxide was moderately successful in explaining the pH dependence of the sorption data on these samples. Two different methods of quantifying the input parameters for the SCM were assessed. While a full SCM for cobalt sorption on these complex environmental substrates is not yet possible, the basic applicability and predictive capability of this type of modeling is demonstrated. A principal requirement to further develop the modeling approach is adequate models for cobalt sorption on component mineral phases of complex environmental sorbents.