Neoproterozoic glacial strata are ubiquitously overlain by thin (<5m) fine-grained carbonate deposits termed "cap
carbonates". These cap carbonates are characterized by anomalously low d13C values (-2 to -6 ). The negative d13C values of the cap carbonates from ca. 850 Ma to ca. 580 Ma, and have provided a useful means of stratigraphic correlation. The stratigraphic position, sedimentology and isotopic composition of the cap carbonates suggest that their occurrence is causally related to glacial climate cycle. A current hypothesis is that the cap carbonate were formed when deep ocean water with low d13C values upwelled and precipitated carbonate on shelves and in cratonic basins. In this hypothesis, the surface-to-deep carbon isotope gradient (Dd13C) would be of much greater magnitude than found in the current oceans (ca. 1.2). We have explored the circulation, redox and nutrient conditions necessary to develop surface-deep d13C gradients with a coupled dynamic model of oceanic C, O and P cycling. Sinking of particulate organic matter (POM) is necessary to drive surface-deep gradients. We find that the Dd13C is mostly independent of surface-deep water exchange rates at steady state, but the rate of approach to steady state does depend on this rate. In contrast, Dd13C depends strongly on P availability and utilisation efficiency. Dd13C is dependent on the redox state of the deep ocean primarily through control of P release during diagenesis of POM on the sea floor. High P inputs to the oceans, very high P utilisation efficiency, and efficient diagenetic recycling of P are all required to drive large Dd13C gradients. If large Dd13C gradients were a long term feature of the Neoproterozoic oceans, some of the conventional wisdom about the interpretation of the very high d13C vlaues in other Neoproterozoic carbonates may need to be revised. Current models may overestimate the burial fraction of organic carbon from the isotope data Estimates of the carbon budget of cap carbonates imply that the precipitation of the caps was an important positive feedback on climate during deglaciation.