There are significant ambiguities concerning the accuracy of oxygen isotope fractionation equations commonly used to retrieve aquatic paleotemperatures from the δ18O value of phosphatic skeletal remains. This is mainly due to the lack of a certified phosphate standard value, discrepancies in measured standard values between laboratories, and differences in methodologies used to constrain phosphate–water fractionation. Depending on the equation used, differences in calculated isotopic water temperatures may range from 4 to 8 °C, these sizable uncertainties significantly reducing the effectiveness of the phosphate 18O/16O ratio as a proxy for water temperature. To address this problem, several phosphate–water fractionation equations from the literature have been tested against the well constrained oxygen isotope fractionation between calcium carbonate and water. Temperatures derived from several pairs of present-day (brachiopods) and fossil (ammonites, brachiopods, belemnite) co-existing carbonate-secreting invertebrates and phosphate-secreting vertebrates (fish) are compared to elucidate the most accurate phosphate-water fractionation equation. These temperatures were obtained by considering using a δ18O value close to 21.7‰ (VSMOW) for the reference phosphorite SRM 120c. The temperature difference (ΔT) calculated from various carbonate–water and phosphate–water oxygen isotope fractionation equations consistently show that the proposed phosphate–water temperature scale established using both modern lingulids and sharks, T(°C) = 117.4(± 9.5) − 4.50(± 0.43) ∗ (δ18OPO4 − δ18OH2O), consistently yields temperatures from bioapatites equivalent to those from co-existing carbonates. Compared to other published phosphate-water fractionation equations, this new equation provides the most reliable estimates of aquatic paleotemperatures for bioapatites.