Transport of selenium in groundwater at the Kesterson Reservoir in the Central Valley of California was strongly retarded because of chemical reduction and precipitation mediated by microbial activity. Under such conditions, negative correlations were documented between aqueous Se and Fe 2+ , Mn, and HzS. Locally, the presence of oxidizing species, notably O2 and NO3, suppressed this reduction, permitting Se mobilization in the shallow aquifer. Selenate, the dominant and most oxidized form of Se, was in electrochemical disequilibrium with subordinate concentrations of selenite. Normally slow inorganic reduction rates were accelerated by microbial activity which utilizes oxidized chemical species including selenate as electron donors during the oxidation of organic matter. Two stratified redox barriers to selenium migration were documented beneath Kesterson: an underlying shallow anoxic zone underlying most of the pond bottom, characterized by high organic content and sulfate reduction, and a deeper dynamic front established by localized Oz infiltration from the overlying ponds and Fe 2+ release from aquifer materials. The reducing nature of this deeper aquifer ultimately precludes Se transport to regional groundwater. [Long et al., 1990]. The U.S. Bureau of Reclamation and the Lawrence Berkeley Laboratory, University of California, have conducted an extensive groundwater chemical monitoring program over the last 6 years at the Kesterson Reservoir. Detailed data tabulations and programmatic reports have been issued on an ongoing basis [Lawrence Berkeley Laboratory, 1987, 1988]. The present paper only attempts to summarize the chemical data in these studies and focuses primarily on determining significant geochemical processes controlling Se mobility. The paper is presented in the context of a preceding paper in this series which discusses the hydrologic and geologic framework at Kesterson and conservative solute transport [Benson et al., this issue]. The conclusions reached in this study are important for designing remedial A total of over 150 monitoring wells were chemically sampled either on a one-time or periodic basis at Kesterson. The wells, situated principally along berms within the reservoir area and also at adjacent offsite locations (Figure 1), varied in total depth between 3 and 65 m. The monitoring well completion methods have been previously described [Benson et al., this issue]. Interstitial water in the anoxic bottom sediments of the ponds were also sampled in pits dug adjacent to the pond margins. Prior to sampling, approximately three well bore volumes of water were removed from a well using high-capacity vacuum and centrifugal pumps. Final sample collection was made using a peristaltic pump connected to a 0.22-/am acetate filter. Chemical parameters including p H, Eh, alkalinity, dissolved oxygen (DO), sulfide, Fe z+, and total Fe were determined in the field. Eh measurements were made in a flow-through cell using a platinum electrode which was checked against a Zobell solution and polished immed...