An inverse finite-difference model of the abyssal circulation in the North Atlantic Ocean is developed in order to evaluate the dynamical information contained in measurements of thorium-230 (230Th). The model has a very coarse resolution and is based on lowest order balances for planetary flows. The naturally occurring 230Th differs from more conventional oceanic tracers in several respects, e.g., its production (by 234U radioactive decay) is globally uniform to a good approximation and its removal can be understood in terms of a simple reversible exchange with particles sinking slowly to the seafloor. The time required for 230Th to reach steady state with respect to particle exchange is estimated to increase with depth, reaching O(10) yr below 1000 m. In the North Atlantic 230Th activities at distant locations share a similar increase with depth in the upper 1000m—a pattern consistent with a reversible exchange—but show drastic differences in the abyssal interior. Two inversions are conducted in order to determine whether the 230Th differences reflect the effects of the circulation—by preventing the slow attainment to steady state w.r.t. particle exchange in deep water—and provide complementary information about the abyssal flow. In a first inversion, observations of density from a hydrographic compilation and of volume transports at specific locations are combined with the dynamical balances in order to infer the basin-scale flow. The inferred flow displays the western boundary current and coherent structures in the abyssal interior with low statistical significance. In a second inversion, the flow is further constrained by the 230Th measurements and the condition that 230Th divergence by the flow field and particle sinking must be locally balanced by 230Th production from 234U decay. The addition of 230Th leads to the estimation of a larger amplitude of the integrated meridional transports below 1000 m (by 2–9 Sv), where the range reflects the uncertainties in the large scale 230Th distribution and in the radiochemical balance. This result is interpreted as a correction by 230Th for the tendency of inverse geostrophic models to lead to the inference of a vanishing circulation when horizontal density gradients are insignificant.