Ultraviolet photolysis of sulfur dioxide (SO 2 ) is hypothesized to be the source of the sulfur isotope mass-independent fractionation (S-MIF) observed in Archean sulfate and sulfide minerals and modern stratospheric sulfate aerosols. A series of photochemical experiments were performed to examine the excitation band dependence of S-MIF during the photochemistry of SO 2 under broadband light sources (a xenon arc lamp and a deuterium arc lamp). Optical filters (200±35nm bandpass and 250nm longpass filters) were used to separately access two different excitation bands of SO 2 in the 190–220nm and the 250–330nm absorption regions, respectively.UV irradiation of SO 2 in the 190–220nm and 250–330nm regions both produced elemental sulfur (S 0 ) and sulfur trioxide (SO 3 ) as end products but yielded very different sulfur isotope signatures. The elemental sulfur products from direct photolysis in the 190–220nm region were characterized by high δ 34 S values (154.7–212.0‰), modest Δ 33 S anomalies of 21±3‰, and relatively constant 33 λ (=ln(δ 33 S+1)/ln(δ 34 S+1)) values of 0.64±0.3, all with respect to the initial SO 2 . Photoexcitation in the 250–330nm region produced elemental sulfur with δ 34 S values of 7.7–29.1‰ and Δ 33 S values of 15.0±1.6‰. In both excitation regions, the SO 3 products were mass dependently fractionated relative to the SO 2 reservoir. The two different absorption regions produced contrasting Δ 36 S/Δ 33 S signatures in the elemental sulfur products, with Δ 36 S/Δ 33 S=−1.9±0.3 and 0.64±0.3 for the 190–220nm and 250–330nm bands, respectively.Our results provide several critical constraints on the origin of the S-MIF signatures observed in modern stratospheric aerosols and in the Archean geological record. A lack of S-MIF in the sulfate product and positive Δ 36 S/Δ 33 S ratios for the elemental sulfur from SO 2 photo-oxidation demonstrate that photoexcitation in the 250–330nm region is not a likely source for the S-MIF observed in modern stratospheric aerosols. Large δ 34 S fractionation, 33 λ values, and Δ 36 S/Δ 33 S ratios observed for the 190–220nm band are qualitatively consistent with predictions from synthetic isotopologue-specific cross sections. These isotope patterns, however, are not compatible with the Archean rock record. We explore the possibility that S-MIF from both the 190 to 220nm and the 250 to 330nm absorption bands could have contributed to the Archean S-MIF signatures.