A method for estimating the release of contaminants from contaminated sites under reducing conditions is proposed. The ability of two chemical reducing agents, sodium ascorbate and sodium borohydride, to produce different redox environments in a gold mining soil contaminated with arsenic was investigated. Liquid-solid partitioning experiments were carried out in the presence of each of the reducing agents at different pH conditions. Both the effect of varying concentrations of the reducing agent and the effect of varying pH in the presence of a constant concentration of the reducing agent were studied. Concentrations of sodium ascorbate ranging from 0.0075 to 0.046 mol L-1 and concentrations of sodium borohydride ranging from 0.0075 to 0.075 mol L-1 were examined. The addition of varying concentrations of sodium borohydride provided greater reducing conditions (ranging from -500 to +140mV versus NHE) than that obtained using sodium ascorbate (ranging from -7 to +345 mV versus NHE). The solubilization of arsenic and iron was significantly increased by the addition of sodium ascorbate for all concentrations examined and pH tested, compared to that obtained under oxidizing conditions (as much as three orders of magnitude and four orders of magnitude, respectively, for the addition of 0.046 mol L-1 of sodium ascorbate). In contrast, the alkaline and highly reduced soil conditions obtained with sodium borohydride lead to a lower effect on arsenic solubilization (as much as one order of magnitude for pH values between ca. 7 and 10 and no effect for pH values between ca. 10 and 12) and no effect on iron solubilization for all concentrations examined and pH tested. At similar ORP-pH conditions the results of extraction for arsenic and iron were different for the two reagents used. (c) 2005 Elsevier B.V. All rights reserved.