Résumé : Coastal oceans play an important role in the carbon cycle and are hotspots of ocean primary production and ocean acidification. These coastal regions are strongly influenced by rives, especially in the Arctic. Despite the importance of the riverine delivery of carbon and nutrients, their effect on the Arctic Ocean is still poorly understood due to hostile conditions and the consequently low number of observations. This thesis aims at improving our understanding of the influence of Arctic riverine delivery of carbon and nutrients by using ocean biogeochemical models.The first part of the thesis evaluated the model skills of the ocean biogeochemical model NEMO-PISCES in the Arctic Ocean. By analyzing model results at different horizontal resolutions, the importance of lateral influx from the adjacent oceans for anthropogenic carbon cycle in the Arctic Ocean wasdemonstrated. These results were then used to adjust a previously published data-based estimate of anthropogenic carbon storage in the Arctic Ocean and the corresponding ocean acidification.In the second part, a pan-Arctic observation-based dataset of riverine carbon and nutrient fluxes was created. This dataset was then used to force the ocean biogeochemical model and the river fluxes were quantified. River fluxes have been shown to sustain up to 24% of Arctic Ocean primary production, to reduce the air-sea CO2 uptake by 20%, and to reduce surface ocean acidification seasonally. Eventually, idealized simulations were made to quantify the sensitivity of the Arctic Ocean biogeochemistry to future changes in riverine delivery of carbon and nutrients. Sensitivities are of small magnitude on a pan-Arctic scale, importance in the coastal areas, and the dominant factor close to river mouths.