Diameter is an important structural trait for root systems, with roots of different thickness havings different physiological functions. Root length is a significant index of root function, while root diameter frequency distribution provides useful information. Thus, studies on the characteristics of root diameter frequency distributionenhance our understanding about the root exploitation strategy for soil resources in different plant or forest systems, as well as the interaction between roots and the soil environment, allowing us to model projections. Using the soil coring method, the diameter frequency distribution characteristics of roots ( ≤ 5 mm in diameter) in six forests was determined using different regeneration approaches. The forests included: the nature forest of Castanopsis carlesii (NF), the secondary forest of Castanopsis carlesii through natural regeneration (NR), the secondary forest of Castanopsis carlesii through natural regeneration with anthropogenic promotion (AR), the Castanopsis carlesii plantation (CC), the Pinus massoniana plantation (PM), and the Cunninghamia lanceolata plantation (CL), in Sanming, Fujian Province. Because it is different to differentiate between the roots of different species, the study was conducted at the community level. There were three main conclusions in this study. First, root length density (RLD) and root biomass density (RBD) for all the three diameter ranges ( ≤ 1 mm, ≤ 2 mm, and ≤ 5 mm) decreased from NF, NR, AR, to CC. Out of the three plantations, the highest RLD and RBD wereobtained in CC, while the lowest RLD and RBD was obtained in PM. Second, out of roots with 0-5mm diameter, 87%-98% and 65%-88% were represented by 0-2 mm and 0-1 mm diameter roots, respectively. The roots of NF, CC, and CL had the similar root diameter frequency distributions, with a maximum frequency of 0.7 mm diameter. The root diameter frequency distribution of AR, NR, and PM was also similar, with a maximum frequency of 0.3mm diameter. The root diameter frequency distribution for all six stands was unimodal, and fitted well with the cumulative lognormal distribution function, with the determination coefficient R² being above 0.99 for all stands. Third, the fitting parameters of μ and σ in the cumulative lognormal distribution function represented different foraging strategies for soil resources in different stands. μ and σ were strongly negatively correlated, which showed a trade-off in the fine root resource acquisition strategy. With the increasing tree species diversity, μ tends to be smaller, while σ tends to be larger, reflecting the growing competition for nutrition and water among the roots of different trees species. μ and σ were distinctly different among different plantation tree species, showing differences in the exploitation strategy for soil resources among plantation tree species. In conclusion, the cumulative lognormal distribution function well reflects stand level root diameter frequency distribution, in addition to root exploitation strategies for soil resources.