The oceanic distributions of oxygen, dissolved nutrients and carbon are strongly affected by the production of particulate material near the ocean surface and its subsequent remineralization during sinking or after deposition on the sea-floor. Dissolved nutrient data can thus be used to derive the rate constants of biogeochemical processes responsible for the observed fields using inverse modeling. Here, a global ocean circulation model is presented that exploits the existing large sets of hydrographic, oxygen, nutrient and carbon data and determines rates of export production and vertical particle fluxes that are compatible with the concentration data. The model is fitted to the property concentration data by systematically varying circulation, air-sea fluxes, production and remineralization rates simultaneously. The adjoint method is applied as an efficient tool for the iterative optimization procedure and produces simulated property fields that are in very good agreement with measurements. The globally integrated export flux of particulate organic matter necessary for the realistic reproduction of nutrient observations is significantly larger than export estimates derived from primary productivity maps. Discrepancies are largest in oligotrophic, open-ocean areas, where the model export fluxes are up to a factor 6 higher than values based on primary productivity estimates of Berger [1989]. This model result is in line with a recent investigation in the subtropical North Pacific that also revealed high open-ocean fluxes. Model export production values are in much closer agreement with estimates based on satellite pigment data in most regions except in the Southern Ocean, where the satellite derived export fluxes seem to underestimate the carbon export significantly.