The declining European forest carbon uptake in recent years is challenging the capacity of European forests to remain carbon sinks, while the European Union’s climate policy relies on this large natural carbon sink to meet its climate targets. The decreased forest carbon sink is thought to be caused by a combination of increased harvest, severe droughts, bark beetle infestations, tree mortality, and slow-down in tree growth causing reduced CO2 uptake. Yet several knowledge gaps remain, importantly a lack of a reconciliation and country-scale quantitative comparison of carbon losses from recent forest disturbances. In the past years, several estimates of forest C fluxes and losses over Europe based on remote sensing have been developed, complementing national greenhouse gas inventory (NGHGI) information. However, an in-depth intercomparison between these various approaches is needed to derive common best estimates of recent C losses, as well as define mutual benefits and individual weaknesses of the different data streams.</p> <p>Here we investigate the recent changes in the European forest carbon sink by combining new high-resolution Earth Observation (EO)-based estimates of forest aboveground biomass carbon (AGC), forest disturbance and recovery, EO-based bookkeeping models, atmospheric CO2 inversions, and NGHGI approaches. We find that all datasets agree on a reduction of forest C uptake in 2020–2023 by 17–27 % compared to the average of 1990–2019 (-135 to -233 TgC/yr), with a remaining forest C sink in 2020–2023 in the range of -100 to -193 TgC/yr. Central and boreal European forest C sink losses contribute most substantially to the observed reduction of the net forest C sink over Europe. This increase in C losses is more strongly driven by effects of natural disturbance than increased harvest, where natural disturbance C losses increased 2.6-fold in central and 1.4-fold in boreal Europe in 2020–23 compared to 2010–2019. EO data at high resolution detect disturbances at lower or higher magnitude to forest C losses reported in NGHGI. Depending on resolution of the satellite product and disturbance detection approach, C losses from small-scale disturbances and management may be better detected, as higher spatial resolution allows capturing small scale heterogeneity.</p> <p>Despite overall agreement of different data streams, we find discrepancies of 2–3 fold difference in AGC and carbon losses estimated for each country between NGHGIs and EO-based datasets. We highlight regional differences across the continent and discuss advantages and challenges of the complementary data streams, with a focus on how high resolution EO and NGHGIs can benefit each other.