We analysed 10 years (2008–2017) of continuous eddy covariance (EC) CO2 flux measurements of net ecosystem exchange (NEE) in a young pedunculate oak forest in Croatia. Measured NEE was gap-filled and partitioned into gross primary productivity (GPP) and ecosystem reparation (RECO) using the online tool by Max Planck Institute for Biogeochemistry in Jena, Germany. Annual NEE, GPP, and RECO were correlated with main environmental drivers. Net primary productivity was estimated from EC (NPPEC), as a sum of −NEE and Rh obtained using a constant Rh:RECO ratio, and from independent periodic biometric measurements (NPPBM). For comparing the NPP at the seasonal level, we propose a simple model that aimed at accounting for late-summer and autumn carbon storage in the non-structural carbohydrate pool. Over the study period, Jastrebarsko forest acted as a carbon sink, with an average (±std. dev.) annual NEE of −319 (±94) gC m−2 year−1, GPP of 1594 (±109) gC m−2 year−1, and RECO of 1275 (±94) gC m−2 year−1. Annual NEE showed high inter-annual variability and poor correlation with annual average global radiation, air temperature, and total precipitation, but significant (R2 = 0.501, p = 0.02) correlation with the change in soil water content between May and September. Comparison of annual NPPEC and NPPBM showed a good overall agreement (R2 = 0.463, p = 0.03), although in all years NPPBM was lower than NPPEC, with averages of 680 (±88) gC m−2 year−1 and 819 (±89) gC m−2 year−1, respectively. Lower values of NPPBM indicate that fine roots and grasses contributions to NPP, which were not measured in the study period, could have an important contribution to the overall ecosystem NPP. At a seasonal level, two NPP estimates showed differences in their dynamic, but the application of the proposed model greatly improved the agreement in the second part of the growing season. Further research is needed on the respiration partitioning and mechanisms of carbon allocation.