This is a study based on determining the stable isotope composition of fluid inclusions of cold- and warm-water corals from the Bahamas and Curaçao respectively. Geochemical data were obtained using a new online technique called the ‘Amsterdam Device’ which comprises δ2H and δ18O analyses of fluid inclusions with a continuous-flow crushing device. The first goal of this study is to test the stable isotope variations in the fluid inclusion of the cold-water coral L. pertusa. This study aligns with the first analysis of Feenstra E.J., 2013, which suggested the included fluid is not of seawater composition. Like Feenstra E.J., the average stable isotope composition of the inclusion fluid is higher in δ18O and lower in δ2H in relation to seawater, indicating that the fluid in cold-water corals is not of seawater composition. This is likely due to non-equilibrium fractionation effects of the coral. Kinetic fractionation effects could explain the simultaneous depletion of δ18O and δ2H . However, this can not explain the fact that δ18O is enriched with respect to isotope equilibrium for aragonite. Rayleigh fractionation should be considered as possible mechanism for the isotope variations in fluid inclusions, since the chemical system is open and the trend is similar to the Global Meteoric Water Line. The second goal of this study is to test whether or not the included fluid can be used for paleotemperature determination. For this technique, the ambient seawater composition during time of coral growth (δw ) is needed, however, it has been identified as a limitation of oxygen isotope paleothermometry. Since the data suggests fluid inclusions in L. pertusa are not of seawater composition, the inclusions do not represent (δw) and therefore this technique does not work on this particular coral species. However, the data of M. annularis fluid inclusions suggests no complex fractionation effects and that this fluid is of seawater composition since the δ18O is approximately the same as seawater. Paleotemperature calculations using the oxygen isotope composition of the included water confirm this statement. Therefore, fluid inclusions in M. annularis might be representative for the ambient seawater during the time of coral growth and offer a solution for the identified limitation of oxygen paleothermometry. Therefore, this new technique could in principle be of great value for marine climatic reconstructions. Since this study focuses mainly on paleotemperature determination, average positive hydrogen composition in relation to SMOW is assumed to be of minor importance in relation to oxygen.