Estimates of photosynthetic and respiratory fluxes at large scales is needed to improve our predictions of the current and future global CO2 cycle. Carbonyl sulphide (OCS) is the most abundant sulphur gas in the atmosphere and has been proposed as a new tracer of photosynthesis (GPP), as the uptake of OCS from the atmosphere is dominated by the activity of carbonic anhydrase (CA), an enzyme abundant in leaves that also catalyses CO2 hydration during photosynthesis. But soils also exchange OCS with the atmosphere which complicates the retrieval of GPP from atmospheric budgets. Indeed soils can take up large amounts of OCS from the atmosphere as soil microorganisms also contain CA, and OCS emissions from soils have been reported in agricultural fields or anoxic soils. To date no mechanistic framework exists to describe this exchange of OCS between soils and the atmosphere but empirical results, once upscaled to the global scale, indicate that OCS consumption by soils dominates over production and its contribution to the atmospheric budget is large, at about one third of the OCS uptake by vegetation, with also a large uncertainty. Here, we propose a new mechanistic model of the exchange of OCS between soils and the atmosphere that builds on our knowledge of soil CA activity from CO2 oxygen isotopes. In this model the OCS soil budget is described by a first-order reaction-diffusion-production equation, assuming that the hydrolysis of OCS by CA is total and irreversible. Using this model we are able to explain the observed presence of an optimum temperature for soil OCS uptake and show how this optimum can shift to cooler temperatures in the presence of soil OCS emissions. Our model can also explain the observed optimum with soil moisture content previously described in the literature as a result of diffusional constraints on OCS hydrolysis. These diffusional constraints are also responsible for the response of OCS uptake to soil weight and depth observed previously. In order to simulate the exact OCS uptake rates and patterns observed on several soils collected from a range of biomes, different CA activities had to be evoked in each soil type, coherent with expected physiological levels of CA in soil microbes and with CA activities derived from CO2 isotope exchange measurements, given the differences in affinity of CA for both trace gases. Our model can also be used to help upscale laboratory measurements to the plot or the region. Several suggestions are given for future experiments in order to test the model further and allow a better constraint on the large-scale OCS fluxes from both oxic and anoxic soils.< Réduire