Deforestation in the tropics is often followed by the creation of anthropogenic savannas used for animal husbandry. By discontinuing burning regimes, forests may recolonize the savanna and carbon stocks may recover. However, little is known about the success and speed of tropical forest recovery, while such information is vital for a better quantification of efforts to reduce emissions from deforestation and forest degradation (REDD+) as well as supporting Forest Landscape Restoration (FLR) practices. Therefore, we designed a forest regeneration experiment within a savanna patch in the Mayombe hills (Democratic Republic of Congo), by discontinuing the annual burning regime in an 88 ha exclosure since 2005. 101 permanent inventory plots (40.4 ha) were installed in 2010 and remeasured in 2014. Tree species were classified as savanna or forest specialists. We estimate a forest specialist encroachment rate of 9 stems ha−1 yr−1 and a savanna specialist disappearance rate of 16 stems ha−1 yr−1. Average diameter of forest specialists did not change due to an increasing influx of recruits, while average diameter of savanna trees increased due to decreasing recruitment. Carbon stored by forest specialists increased from 3.12 to 5.60 Mg C ha−1, suggesting a forest carbon recovery rate of 0.62 Mg C ha−1 yr−1. Using the average carbon stock of 19 nearby mature rainforest plots as a reference, we estimate a total forest carbon recovery time of at least 150 years. The Manzonzi exclosure may potentially become an important reference experiment to quantify REDD+ schemes in Central Africa. Furthermore, this natural regeneration experiment demonstrates how carbon sequestration and biodiversity conservation can go hand-in-hand. However, more censuses are needed to better quantify the long-term carbon recovery trajectory within the protected area.