The Messinian evaporites in Nijar Basin, southeastern Spain, consist of coarse-grained primary selenites and laminated fine-grained gypsums. These gypsums, which are similar to those deposited in other Mediterranean Messinian basins, have rather variable delta S-34 and delta O-18 (SO4) (+21 to +24 parts per thousand for delta S-34 + 10 + 17 parts per thousand for delta O-18) that are generally greater than the normal Tertiary marine values (21-22 parts per thousand for delta S-34 and 13 parts per thousand for delta O-18). Moreover, the variation of delta O-18 (SO4) is larger than that of delta S-34. Continental input and reservoir effect did not play important roles for these isotope variations because these factors should have resulted in lower delta S-34 and delta O-18 (SO4) than the normal marine values. Redox reactions of sulfur species in brine should have been responsible for the isotope variations. To interpret the variable delta S-34 and delta O-18 (SO4), we quantitatively modeled delta S-34 and delta O-18 (SO4) of dissolved sulfate during redox reactions in two marine evaporative settings: in sediment pores in a shallow-water pan, and in a free-brine column in a deep-water basin. Modeled delta S-34 and delta O-18 of dissolved sulfate during reduction of sulfate yield small variations, especially for O isotopes in a shallow-water pan. However, modeled delta S-34 and delta O-18 of dissolved sulfate during reduction and reoxidation cycling processes in a deep-water basin show a 514 progressive increase of up to 10 parts per thousand for both delta S-34 and delta O-18 (SO4) above their normal marine values. Importantly, the modeled pathways for the deep-water basin setting are consistent with the Nijar data, whereas those for pore waters in shallow-water pans are a poor fit to the Nijar data. The modeled results consistent with trace elements, Sr isotope data, and selenite fabrics suggest that these Yesares selenites have formed in deep marine brines. The modeling results also demonstrate that delta O-18 (SO4) can have larger variation than delta S-34 because the incorporation of water oxygen and probably dissolved free oxygen increases delta O-18 of the reoxidized sulfate, whereas reoxidation of sulfide to sulfate decreases the total delta S-34. (SO4). The modeled results could provide important geological implications for isotope and environment interpretations for ancient evaporites. For example, redox reactions of sulfur species during sulfate precipitation could modify normal marine delta S-34 and delta O-18(SO4) more efficiently than freshwater contributions and reservoir effects. Therefore, to establish delta S-34 and delta O-18 (SO4) age curves, redox processes for ancient evaporites should be considered. Copyright (C) 2001 Elsevier Science Ltd.