Sulfur trioxide (SO 3 ) is an unstable pollutant, and its removal from the gas phase of industrial flue gas remains a significant challenge. Herein, we propose a reverse conversion treatment (RCT) strategy to reduce S(VI) in SO 3 to S(IV) by combining bench-scale experiments and theoretical studies. We first demonstrated that metastable sulfides can break the S−O bond in SO 3 , leading to the re-formation of sulfur dioxide (SO 2 ). The RCT performance varied between mono-and binary-metal sulfides, and metastable CuS had a high SO 3 conversion efficiency in the temperature range of 200−300 °C. Accordingly, the introduction of selenium (Se) lowered the electronegativity of the CuS host and enhanced its reducibility to SO 3 . Among the CuSe 1−x S x composites, CuSe 0.3 S 0.7 was the optimal RCT material and reached a SO 2 yield of 6.25 mmol/g in 120 min. The low-valence state of selenium (Se 2− /Se 1− ) exhibited a higher reduction activity for SO 3 than did S 2− /S 1− ; however, excessive Se doping degraded the SO 3 conversion owing to the re-oxidation of SO 2 by the generated SeO 3 2− . The density functional theory calculations verified the stronger SO 3 adsorption performance (E ads = −2.76 eV) and lower S−O bond breaking energy (E a = 1.34 eV) over CuSe 0.3 S 0.7 compared to those over CuS and CuSe. Thus, CuSe 1−x S x can serve as a model material and the RCT strategy can make use of field temperature conditions in nonferrous smelters for SO 3 emission control.