Cl isotope ratios measured on HCl thermally evolved from as-yet-unknown phases in sedimentary rocks and sand in Gale Crater provide unexpected insights to the Martian surficial Cl cycle. The seven samples yield δ37Cl values ranging from −1±25‰ to −51±5‰. Five analyses from two samples of the Sheepbed mudstone (Yellowknife Bay study area) are analytically indistinguishable with a mean δ37Cl of −11±7‰(1σ). In contrast, four mudstones/sandstones from the Kimberley and Pahrump study areas also yielded indistinguishable ratios, but with a mean δ37Cl of −43±6‰. The Rocknest sand deposit gave a highly uncertain δ37Cl value of −7±44‰.These light and highly variable δ37Cl values are unique among known solar system materials. Two endmember models are offered to account for these observations, and in both, perchlorate, with its extreme ability to fractionate Cl isotopes, is critical. In the first model, SAM is detecting HCl from an oxychlorine compound (e.g., perchlorate) produced from volcanic gas emissions by atmospheric chemical reactions. Similar reactions in Earth's atmosphere may be responsible for the isotopically lightest known Cl outside of this study, in perchlorate from the Atacama Desert. Some of the Gale Crater δ37Cl values are more negative than those in Atacama perchlorate, but because reaction mechanisms and associated fractionation factors are unknown, it is impossible to assess whether this difference is prohibitive. If the negative δ37Cl signal is produced in this fashion, the isotopic variability among samples could arise either from variations in the relative size of the reactant chloride and product perchlorate reservoirs, or from variations in the fraction of perchlorate reduced back to chloride after deposition. Such reduction strongly enriches 37Cl in the residual perchlorate.Perchlorate reduction alone offers an alternative endmember model that can explain the observed data if SAM measured HCl derived from chloride. In this model isotopically normal perchlorate produced by an unspecified mechanism is reduced to chloride. Depending on the relative size of the reduced reservoir, the integrated product chloride can vary in isotopic composition from −70‰ in the first increment all the way to the starting composition if the perchlorate is fully reduced. Thus, variable degrees of perchlorate reduction can produce chloride with the appropriate δ37Cl range. Combination of the two endmember models, in which the perchlorate subject to post-deposition reduction is isotopically negative from atmospheric reactions, is also possible.Determination of the phase hosting the Cl measured by SAM, an oxychlorine compound or chloride, is critical for selecting between these models, and for developing implications of the results for the Mars surficial Cl cycle. At present it is not possible to conclusively establish which phase is responsible (possibly both), but limited evidence favors the conclusion that the measured Cl derives mostly from an oxychlorine compound.