Adsorption processes are extensively used for the selective removal of trace elements from drinking water. They must be designed not only with equilibrium data but also with mass-transfer data. Indeed, this paper investigates the transport of arsenate in natural manganese oxide columns. Column experiments were run at different conditions of particle sizes and flow rates, including an interrupted-flow experiment. The breakthrough curve analysis showed that transport was affected by nonlinear adsorption and intraparticle diffusion. Nonconventional features were observed, such as the variation of the total adsorption capacity with the flow rate and with the particle size. They were attributed to the complex porous structure of the grains. Results were interpreted by means of a transport model including Langmuir adsorption and mass transfer, with a single adjustable parameter: the effective diffusivity of arsenate in the grain. This parameter included both adsorption and diffusion. In a second step, diffusivities including intraparticle diffusion alone were calculated. Values between 0.6 and 7.0 x 10(-11) m(2) s(-1) were found; they were close to published data of the pore diffusivity of arsenate in activated alumina grains. Moreover, the model, despite its simplicity, succeeded in predicting the breakthrough points of arsenate at different flow rates with a single value of the effective diffusivity. This modeling approach is of great importance for the design of adsorption processes for arsenate removal from drinking water.