Biological production lowers the CO 2 concentrations in the surface layer of the ocean, and sinking detritus ''pumps'' nutrients and CO 2 into the deep ocean. Quantifying the efficiency of the biological pump is a prerequisite for global CO 2 budgets. Sediment traps are commonly used to directly measure the vertical particle flux; however, for logistical and financial reasons, traps cannot provide area-wide data sets. Moreover, it has been shown that sediment traps can under-or overestimate particle fluxes considerably. In this paper, we present a new technique to estimate the downward flux of particulate matter with an adjoint model. Hydrographic and nutrient data are used to calculate the mean ocean circulation together with parameters for particle fluxes using the AWI Adjoint Model for Oceanic Carbon Cycling (AAMOCC). The model is fitted to the property concentrations by systematically varying circulation, air -sea fluxes, export production and remineralization rates of particulate biogenic matter simultaneously. The deviations of model fluxes based on nutrient budgets from direct measurements with sediment traps yield an independent estimate of apparent trapping efficiencies. While consistent with hydrographic and nutrient data, model particle fluxes rarely agree with sediment trap data: (1) At shallow water depth ( V 1000 m), sediment trap fluxes are at the average 50% lower than model fluxes, which confirms flux calibrations using radionuclides; (2) in the very deep traps, model fluxes tend to be lower compared to data, which might be explained by lateral inputs into the traps. According to these model results, particle fluxes from the euphotic zone (EP) into mid water depth are considerably higher and the shallow loop of nutrient is more vigorous than would be derived from sediment trap data. Our results imply that fluxes as collected with sediment traps are inconsistent with model derived long-term mean particle fluxes based on nutrient budgets in the water column. In agreement with recent radionuclide studies, we conclude that reliable export flux estimates can only be obtained from sediment trap data if appropriate corrections are applied. D