1. Due to intensifying human disturbance, over half of the world’s tropical forests are reforested or afforested secondary forests or plantations. Understanding the resilience of carbon (C) stocks in these forests, and estimating the extent to which they can provide equivalent carbon (C) sequestration and stabilization to the old growth forest they replace, is critical for the global C balance. 2. In this study, we combined estimates of biomass C stocks with a detailed assessment of soil C pools in bare land, Eucalyptus plantation, secondary forest, and natural old-growth forest after over 50 years of forest restoration in a degraded tropical region of South China. We used isotope studies, density fractionation and physical fractionation to determine the age and stability of soil C pools at different soil depths. 3. After 52 years, the secondary forests had equivalent biomass C stocks to natural forest, whereas soil C stocks were still much higher in natural forest (97.42 t ha-1) than in secondary forest (58.75 t ha-1) or Eucalyptus plantation (38.99 t ha-1) and lowest in bare land (19.9 t ha-1). Analysis of δ13C values revealed that most of the C in the soil surface horizons in the secondary forest was new C, with a limited increase of more recalcitrant old C, and limited accumulation of C in deeper soil horizons. However, occlusion of C in microaggregates in the surface soil layer was similar across forested sites, which suggests that there is great potential for additional soil C sequestration and stabilization in the secondary forest and Eucalyptus plantation. 4. Collectively, our results demonstrate that reforestation on degraded tropical land can restore biomass C and surface soil C stocks within a few decades, but much longer recovery times are needed to restore recalcitrant C pools and C stocks at depth. Repeated harvesting and disturbance in rotation plantations had a substantial negative impact on the recovery of soil C stocks. We suggest that current calculations of soil C in secondary tropical forests (e.g. IPCC Guidelines for National Greenhouse Gas Inventories) could overestimate soil C sequestration and stabilization levels in secondary forests and plantations.