An 18 million year record of the Ca isotopic composition (d44/42Ca) of planktonic foraminiferans from ODP site 925, in the Atlantic, on the Ceara Rise, provides the opportunity for critical analysis of Ca isotope-based reconstructions of the Ca cycle. d44/42Ca in this record averages +0.37 ± 0.05 (1r SD) and ranges from +0.21‰ to +0.52‰. The record is a good match to previously published Neogene Ca isotope records based on foraminiferans, but is not similar to the record based on bulk carbonates, which has values that are as much as 0.25‰ lower. Bulk carbonate and planktonic foraminiferans from core tops differ slightly in their d44/42Ca (i.e., by 0.06 ± 0.06‰ (n = 5)), while the difference between bulk carbonate and foraminiferan values further back in time is markedly larger, leaving open the question of the cause of the difference. Modeling the global Ca cycle from downcore variations in d44/42Ca by assuming fixed values for the isotopic composition of weathering inputs (d44/42Caw) and for isotope fractionation associated with the production of carbonate sediments (Dsed) results in unrealistically large variations in the total mass of Ca2+ in the oceans over the Neogene. Alternatively, variations of ±0.05‰ in the Ca isotope composition of weathering inputs or in the extent of fractionation of Ca isotopes during calcareous sediment formation could entirely account for variations in the Ca isotopic composition of marine carbonates. Ca isotope fractionation during continental weathering, such as has been recently observed, could easily result in variations in d44/42Caw of a few tenths of permil. Likewise a difference in the fractionation factors associated with aragonite versus calcite formation could drive shifts in Dsed of tenths of permil with shifts in the relative output of calcite and aragonite from the ocean. Until better constraints on variations in d44/42Caw and Dsed have been established, modeling the Ca2+ content of seawater from Ca isotope curves should be approached cautiously.