Abstract. The decoupled behaviour observed between Nd isotopic composition (Nd IC, also referred as ε_Nd) and Nd concentration cycles has led to the notion of a "Nd paradox". While ε_Nd behaves in a quasi-conservative way in the open ocean, leading to its broad use as a water-mass tracer, Nd concentration displays vertical profiles that increase with depth, together with a deep-water enrichment along the global thermohaline circulation. This non-conservative behaviour is typical of nutrients affected by scavenging in surface waters and remineralisation at depth. In addition, recent studies suggest the only way to reconcile both concentration and Nd IC oceanic budgets, is to invoke a "Boundary Exchange" process (BE, defined as the co-occurrence of transfer of elements from the margin to the sea with removal of elements from the sea by Boundary Scavenging) as a source-sink term. However, these studies do not simulate the input/output fluxes of Nd to the ocean, and therefore prevents from crucial information that limits our understanding of Nd decoupling. To investigate this paradox on a global scale, this study uses for the first time a fully prognostic coupled dynamical/biogeochemical model with an explicit representation of Nd sources and sinks to simulate the Nd oceanic cycle. Sources considered include dissolved river fluxes, atmospheric dusts and margin sediment re-dissolution. Sinks are scavenging by settling particles. This model simulates the global features of the Nd oceanic cycle well, and produces a realistic distribution of Nd concentration (correct order of magnitude, increase with depth and along the conveyor belt, 65% of the simulated values fit in the ±10 pmol/kg envelop when compared to the data) and isotopic composition (inter-basin gradient, characterization of the main water-masses, more than 70% of the simulated values fit in the ±3 ε_Nd envelop when compared to the data), though a slight overestimation of Nd concentrations in the deep Pacific Ocean may reveal an underestimation of the particle fields by the biogeochemical model. Our results indicate 1) vertical cycling (scavenging/remineralisation) is absolutely necessary to simulate both concentration and ε_Nd, and 2) BE is the dominant Nd source to the ocean. The estimated BE flux (1.1×10¹⁰ g(Nd)/yr) is much higher than both dissolved river discharge (2.6×10⁸ g(Nd)/yr) and atmospheric inputs (1.0×10⁸ g(Nd)/yr) that both play negligible role in the water column but are necessary to reconcile Nd IC in surface and subsurface waters. This leads to a new calculated residence time of 360 yrs for Nd in the ocean. The BE flux requires the dissolution of 3 to 5% of the annual flux of continental weathering deposited via the solid river discharge to the continental margin.