Water vapor and carbon dioxide are the main atmospheric constituents which are controlling the Earth’s climate. The rapid increase in the atmospheric content of carbon dioxide since the beginning of the industrial revolution is considered as one of the main drivers of the recent climate changes on Earth. Only about 40% of the total anthropogenic emissions of CO2 remain in the atmosphere, while the rest of the emitted CO2 is stored in oceans and land. The terrestrial sinks of carbon are global soils and forests. Forests sequester CO2 from the atmosphere and assimilate carbon into above- and below ground biomass and by that partially offset anthropogenic emission of CO2 and participate in a regulation of climate. Because of these findings, monitoring of CO2 exchange between atmosphere and underlying forest ecosystems has gained significant importance. Micrometeorological eddy covariance technique has shown as the most accurate way for direct flux measurement of trace gases, and today it is a standard tool for estimating net ecosystem exchange (NEE) of trace gases between the atmosphere and the underlying surface. Within this research 10-year eddy covariance experiment (2008-2017) was carried out in young pedunculate oak (Quercus robur L.) stands (35-44 years old) which are part of the forest complex of the Kupa River basin, about 35 km SW from Zagreb, Croatia. Over the entire study period, Jastrebarsko forest acted as a carbon sink, with an average annual NEE of -319 ± 30 g C m-2 yr-1. Estimated NEE was partitioned into gross primary production (GPP) and ecosystem respiration (RECO). Furthermore, RECO was partitioned into heterotrophic (Rh) and autotrophic respiration (Ra). Most important carbon flux in forest ecosystems, net primary production (NPP), was estimated by subtracting NEE from heterotrophic respiration. In this study the causes of inter-annual variability of carbon NEE were investigated. Also, the impact of extreme weather events (droughts and floods) on carbon fluxes was investigated. For validation of EC measurements, a biometric estimate of the net primary productivity (NPPBM), which was built on periodic measurement and simple modelling, was compared with NPPEC. Comparison of NPPEC and NPPBM showed good agreement (R2=0.46).