Fine roots, commonly defined as roots <2 mm in diameter, play an important role in nutrient cycle and energy flow in forest ecosystems.Tree fine roots are highly spatially heterogeneous, so only a large number of samples can be valid to study such heterogeneity.However, the minimum sample numbers and the proper sampling strategy is unknown yet, which results in uncertainties in fine-root studies.In this paper, by using the soil core sampling method, 96 soil cores were taken randomly in a subtropical evergreen broad-leaved forest located at Sanming, Fujian Province.Every soil core was divided into four parts according to soil depth: 0-10 cm, 10-20 cm, 20-40 cm, and 40-60 cm.Fine roots in different soil depth were washed and picked, and was further separated into different diameter classes of 0-1 mm and 1-2 mm.Fine roots were then scanned for measurement of morphological characteristics, and oven and weighed for biomass.The objectives of this research were to: (1) understand the spatial heterogeneity of fine root biomass and morphological characteristics; (2) estimate the sample sizes required for getting various fine root characteristic values.The results showed that, as diameter class increases, the coefficient of variation (CV) of fine root variables increases, and the required sampling sizes increase accordingly.The CVs for fine-root biomass, length and surface area per unit ground area of different diameter classes were unanimously between 40%-60%.At the same time, fine-root biomass, length and surface area at per unit ground area declined with the increases of fine root diameter class.With the increases of soil depth, fine-root biomass of different diameter classes per unit ground area decreased, and their CVs increased, while the required sampling sizes increased accordingly.Except for fine-root biomass of 0-1 mm per unit ground area in the surface soil which had a CV of 48.61%, the rest of fine roots from different diameter classes and soil depths had CVs of biomass per unit ground area consistently between 80%-90%.Under the condition with confidence leve1 of 98% and precision of 80%, for fine roots of 0-1 mm and 1-2 mm, collecting 16 and 42 cores can meet the determination for fine root biomass per unit ground area, 17 and 31 cores for fine root length per unit ground area, and 25 and 33 cores for fine root surface area per unit ground area, respectively.By comparison, most reported root studies using soil coring were insufficient sampling sizes.The Shapiro-Wilk test showed that among different diameter classes, only fine-root biomass per unit ground area of 0-1 mm conforms to the normal distribution; among different soil depths, only fine-root biomass per unit ground area of 0-1 mm in the surface soil conforms to normal distribution.Thus, for field data on fine-root biomass and morphology by soil coring, data transformation were generally required to meet normality.These results provided a scientific basis for minimum sampling sizes and sampling strategy in subtropical evergreen broad-leaved forest and are therefore helpful to belowground studies.More studies should be conducted to further determine factors affecting the minimum sampling size of soil coring.