The thermal dehydration of gypsum (CaSO4 ·2H2O) was studied via the novel method of differential thermal isotope analysis (DTIA) combined with in situ Raman spectroscopy and TG/DSC measurements. The objective was to determine whether an intermediate hemihydrate phase (CaSO4 ·nH2O) exists under differing dehydration conditions, by varying the heating rate, the partial pressures of water (P H2O ) and the gypsum size fraction (grain diameters ~20μm to 2mm). Here we definitively show that the occurrence of the intermediate bassanite (CaSO4 ·0.5H 2 O) phase is primarily dependent on grain size. This contrasts with previous studies that state that the dehydration behaviour of gypsum is solely dependent on heating rate, P H2O , or that it simply occurs under all dehydration conditions. We show that bassanite becomes undetectable by Raman spectroscopy, Thermal gravimetric Analysis (TGA) and Differential Scanning Calorimetry (DSC) during the dehydration of gypsum crystals with large (>500μm) grain sizes. Its occurrence increases systematically with decreasing grain size, while lower heating rates and higher P H2O stabilize this phase for a longer length of time. This is confirmed by DTIA measurements that show, during the dehydration of powdered gypsum (~200μm) to bassanite and subsequently to anhydrite, there is differential fractionation of oxygen and hydrogen isotopes between the two dehydration steps. At larger (>500μm) grain sizes, no change in isotopic composition is observed. Results also confirm that the water molecules within the gypsum structure are equivalent and that two-step nature of the dehydration process is caused by kinetic factors. These findings consolidate the discrepancies in the large volume of literature regarding the gypsum-bassanite- anhydrite transition.