Arsenic is often assumed to be immobile in sulfidic environments. Here, laboratory-scale microcosms were conducted to investigate whether microbial sulfate reduction could control dissolved arsenic concentrations sufficiently for use in groundwater remediation. Sediments from the Vineland Superfund site and the Coeur d'Alene mining district were amended with different combination of lactate and sulfate and incubated for 30-40 days. In general, sulfate reduction in Vineland sediments resulted in transient and incomplete arsenic removal, or arsenic release from sediments. Sulfate reduction in the Coeur d'Alene sediments was more effective at removing arsenic from solution than the Vineland sediments, probably by arsenic substitution and adsorption within iron sulfides. X-ray absorption spectroscopy indicated that the Vineland sediments initially contained abundant reactive ferrihydrite, and underwent extensive sulfur cycling during incubation. As a result, arsenic in the Vineland sediments could not be effectively converted to immobile arsenic-bearing sulfides, but instead a part of the arsenic was probably converted to soluble thioarsenates. These results suggest that coupling between the iron and sulfur redox cycles must be fully understood for in situ arsenic immobilization by sulfate reduction to be successful.