Background and Objectives: Forest encroachment remains a critical issue in Thailand relating to anthropogenic disturbances, leading to the degradation of plant community structures and biodiversity loss. Nevertheless, these disturbed forests still retain the potential for natural regeneration, particularly in areas where remnant adult trees persist and serve as seed sources for the recolonization of native species with pioneer species during successional process. This study aimed to compare the forest structure and the amount of carbon storage in naturally forest restoration areas after disturbances. The research was conducted at the Wang Nam Khiao Forestry Student Training and Research Station, Nakhon Ratchasima Province, northeastern Thailand. The findings are intended to provide essential baseline data for planning on forest ecosystem restoration and enhancing long-term carbon sequestration capacity in a sustainable manner. Methodology: A permanent sample plot or forest dynamics plot (FDP) of 100 m × 100 m was established in each of three forest community types undergoing natural recovery after disturbances: restored deciduous dipterocarp forest (RDDF), restored mixed deciduous forest restoration (RMDF), and restored dry evergreen forest restoration (RDEF), respectively. The FDP is useful for long-term ecological research, LTER. Then, each plot was subdivided into 100 subplots of 10 m × 10 m. Within these subplots, all trees with a diameter at breast height (DBH) of at least 1 cm were tagged, measured the DBH and total tree height, and identified species. Unidentified tree species were specimen collected, then, compared with identified species at Bangkok Herbarium, Department of National Parks, Wildlife and Plants Conservation. The data was then analyzed to determine the importance value index (IVI) of each species, species diversity index followed by Shannon-Weiner (H’), tree distribution form based on DBH size class distribution, and carbon storage based on aboveground biomass (ABG). Statistical test of quantitative values among forest types was applied using one-way ANOVA and Tukey's HSD at p < 0.05. Main Results: We found that the restored dry evergreen forest restoration, DEFR, exhibited the most prominent characteristics in terms of forest structure and biodiversity among restored forest types, with a total of 86 tree species recorded. This was followed by the restored mixed deciduous forest, RMDF, with 76 species and the restored deciduous dipterocarp forest, RDDF, with 45 species, respectively. These patterns corresponded closely with measurements of tree density and basal area, which were also highest in the RDEF (2,296 stems ha-1 and 30.66 m² ha-1, respectively), followed by the RMDF (1,735 stems ha-1 and 21.35 m² ha-1) and lowest in the RDDF (800 stems ha-1 and 19.13 m² ha-1). Similarly, the Shannon–Wiener diversity index (H′) revealed statistically significant differences among restored forest types (p < 0.001), with the RDEF showing the highest diversity (H′ = 3.63), followed by the RMDF (H′ = 2.98) and the RDDF (H′ = 2.30), respectively. The analysis of tree size class distribution based on diameter at breast height (DBH) indicated that both the RMDF and RDEF exhibited a negative exponential form (L-shape), suggesting ongoing regeneration and the ability to maintain their forest structure in the future. The environmental factors such as soil properties and available water available in these types are suitable for tree species establishment. In contrast, the RDDF displayed a unimodal form or bell-shape, reflecting a discontinuity in the size class transition, particular a lack of smaller-sized individuals. This pattern reflects a regeneration constraint in the RDDF, likely resulting from more frequent and severe wildfires compared to the other restored forest types. Carbon storage based on ABG also varied significantly among restored forest types (p < 0.01). The RDEF had stored the highest amount of carbon storage (91.15 t C ha-1), followed by the RDDF (50.92 t C ha-1) and the RMDF (31.22 t C ha-1). These findings emphasize the important of forest structure—particularly tree density and basal area—in enhancing carbon sequestration capacity. Interestingly, although the RMDF had higher species diversity and basal area than the RDDF, it stored less carbon. This is likely attributable to species composition, as the RMDF was dominated by pioneer or fast-growing species such as Microcos tomentosa, Wrightia arborea, and Millettia leucantha, which typically accumulate less biomass. In contrast, the RDDF was composed native or climax species, including Shorea siamensis and Dipterocarpus intricatus, which contribute more substantially to carbon storage. Conclusion: Among the restoration forests, the RDEF exhibited a relatively well-developed plant community structure, both in terms of biodiversity and carbon sequestration, when compared to the RMDF and RDDF. Forest structure and species composition played a crucial role in determining of carbon storage potential. Therefore, forest restoration efforts should consider enhancing species diversity during the early successional stages by integrating planting of pioneer species with native late-successional species that related to their niche with the degraded areas. They will facilitate the suitable environments for climax species, then, recovery into the previous forests. This approach can support the recovery of natural forest conditions while simultaneously promoting efficient carbon sequestration during the plant successional process.