The kinetics of As(V) adsorption by ferrihydrite was investigated in coprecipitation and postsynthesis adsorption experiments conducted in the pH range 7.5-9.0. In coprecipitation experiments, As(V) was present in solution during the hydrolysis and precipitation of iron. In adsorption experiments, a period of rapid (<5 min) As(V) uptake from solution was followed by continued uptake for at least eight days, as As(V) diffused to adsorption sites on ferrihydrite surfaces within aggregates of colloidal particles. The time dependence of As(V) adsorption is well described by a general model for diffusion into a sphere if a subset of surface sites located near the exterior of aggregates is assumed to attain adsorptive equilibrium rapidly. The kinetics of As(V) desorption after an increase in pH were also consistent with diffusion as a rate-limiting process. Aging of pure ferrihydrite prior to As(V) adsorption caused a decrease in adsorption sites on the precipitate owing to crystallite growth. In coprecipitation experiments, the initial As(V) uptake was significantly greater than in post-synthesis adsorption experiments, and the rate of uptake was not diffusion limited because As(V) was coordinated by surface sites before crystallite growth and coagulation processes could proceed. After the initial adsorption, As(V) was slowly released from coprecipitates for at least one month, as crystallite growth caused desorption of As(V). Adsorption densities as high as 0.7 mole As(V) per mole of Fe were measured in coprecipitates, in comparison to 0.25 mole As(V) per mole of Fe in post-synthesis adsorption experiments. Despite the high Concentration of As(V) in the precipitates, EXAFS spectroscopy ( WAYCHUNAS et al., 1993) showed that neither ferric arsenate nor any other As-bearing surface precipitate or solid solution was formed. The high adsorption densities are possible because the ferrihydrite particles are extremely small, approaching the size of small dioctahedral chains at the highest As(V) adsorption density. The results suggest that the solid solution model proposed by Fox (1989, 1992) for control of arsenate and phosphate concentrations in natural waters may be invalid.