Crystalline calcium carbonate isotope compositions have been widely used to reconstruct past environments. However, if their isotopic compositions are modified because of crystallization from an amorphous precursor, their reliability as paleo-geochemical proxies can be compromised. This study explored the changes in the oxygen isotope compositions during the transformation of amorphous calcium carbonate (ACC) into crystalline carbonate under different conditions of relative humidity (RH of 33–95 %), temperature (T of 5 °C and 20 °C) and in the presence/absence of atmospheric CO2. The data showed that at low RH and T (e.g., RH ≤ 45 % and 5 °C) when a complete ACC-crystalline carbonate transformation did not take place then the original ACC δ18O values (δ18OCaCO3 = −15.9 ± 1.0 ‰, VPDB) were preserved throughout the experimental runtime (up to 144 days). In contrast, in fully crystallized CaCO3 (e.g., at RH ≥ 60 %) the δ18OCaCO3 values increased rapidly over the first few days, followed by a slower and gradual increase. By the end of the experiments (i.e., after 103–144 days) the crystalline δ18OCaCO3 values ranged from −10.4 ‰ to −8.1 ‰ in the presence of atmospheric CO2 and from −12.6 ‰ to −9.5 ‰ in the CO2-free experiments. These changes in oxygen isotope compositions of the CaCO3 reaction products (calcite and/or vaterite) were mainly driven by exchange with H2O from the hydrated ACC i.e. the synthesis fluid. In CO2-present experiments, oxygen isotope fractionation factors between the CaCO3 reaction products and the synthesis fluid (18αc–w) approached or exceeded oxygen isotope equilibrium values. This could be explained by a decrease in the initially high pH of the aqueous fluid released from ACC dissolution during CO2 hydration/hydroxylation, which would have increased the oxygen isotope exchange kinetics between H2O and dissolved inorganic carbon (DIC). In some experiments, the hydration/hydroxylation of 18O-enriched CO2, due to isotopic salt-effects, might have also resulted in 18O-enriched calcium carbonates and calculated fractionation factors that exceeded equilibrium values. In the CO2-free experiments, isotopic equilibrium between the crystalline phase and the synthesis fluid was not reached. This oxygen isotope disequilibrium suggests that without the pH lowering effect of the hydroxylation/hydration of CO2, the CO32− released during ACC/calcite dissolution-reprecipitation may have not isotopically equilibrated with the high pH synthesis fluid due to the long equilibration times required to reach isotope equilibrium at high pH values, leading to the self-buffering of δ18OCaCO3 values. The results suggest that the oxygen isotopic compositions of natural carbonates formed from ACC transformation in air and at low water/solid ratio (e.g., biominerals or carbonates formed in caves) are complex and cannot be used as simple proxies if the reaction kinetics (RH/CO2/T) and H2O sources are not known and quantified.