This study was conducted in Larix principis-rupprechtii forests, which located in Guandi Mountain, Shanxi Province, China. Two plots(32m×32m) were placed. Plot A was disturbed stand while plot B undisturbed. Basic data of fine root biomass with various diameters were collected from three depths(0-10cm, 10-20cm and 20-30cm) of soil layers via root drilling methods. The fine roots were separated into live and dead fine root (≤2mm). The live fine roots were classified into three categories (≤1mm, 1-2mm, 2-5mm). The spatial heterogeneity of fine root biomass was analyzed based on theory and methodology of spatial pattern analysis in geostatistics. By comparing the fine root biomass at same soil depth in different stand, fine root biomass was higher in undisturbed stand (plot B) with great difference, which also has high spatial heterogeneity. After cutting disturbance, fine root biomass became lower with smaller difference and the spatial heterogeneity intensity of fine root biomass became weaker. In different soil layer, the heterogeneity of fine root biomass changed significantly (P<0.05). Results showed: (1) At 0-10cm soil depth, fine root(≤1mm, 1-2mm, 2-5mm and ≤2mm dead root ) biomass decreased respectively 8.14%, 48.29%, 47.96% and 74.21%. Most variation of fine root biomass was spatially explained in undisturbed stand. Spatial heterogeneity of fine root biomass performed clearly randomly variation after disturbance. Spatial distribution of fine root (≤1mm)biomass became more complicated in disturbed stand. (2) At 10-20cm soil depth, fine root(1-2mm, 2-5mm and ≤2mm dead root ) biomass decreased respectively 60.53%,67.56%,67.32%。Fine root biomass showed weaker spatial heterogeneity intensity in disturbed stand, which was only 5.4%-88.9% spatial heterogeneity in undisturbed stand. (3) At 20-30cm soil depth, fine root(1-2mm, 2-5mm and ≤2mm dead root ) biomass decreased respectively 127.90%, 44.22%, 355.19%. In both plot A and B, fine biomass of all classes changed significantly (P<0.05). Fine root biomass(≤1mm) remained mainly spatial autocorrelation variation, with structural variance proportion 86.1% at the range of 0.5-2.9m. After cutting disturbance, spatial heterogeneity became weaker and performed clearly randomly variation. Vertical differentiation appeared that spatial heterogeneity of fine root biomass decreased with the increasing of soil depths in two stands.