To cross-validate the accuracy of carbon flux estimates of the eddy covariance with that of the stand inventory method in temperate forests, we investigated the spatial variations in forest biomass and their driving factors within the fetch (1500 m×400 m) of the Maoershan flux tower using a geostatistics approach and regression methods.One hundred and six circular plots with diameters of 20 m were established by a 100 m×50 m grid using a total station. All of the trees with diameter at breast height (DBH) greater than 2 cm were measured in each plot. The biomass and its components were calculated based on the DBH data and the site-species-specific biomass equations developed previously. The ranges of the total, above-, and below-ground biomasses were 22.68-304.89, 17.99-245.96, and 4.69-69.25 Mg/hm², respectively, with corresponding means of 155.64, 124.17, and 31.47 Mg/hm². The root:shoot ratio (RSR) ranged between 0.18 and 0.36, with a mean of 0.25. For the total biomass, the contribution of specific species was ranked in the following order:Ulmus japonica (22.80%) > Fraxinus mandshurica (15.70%) > Betula platyphylla (15.52%) > Betula costata (7.23%) > Juglans mandshurica (6.21%) > Acer mono (6.20%), with the remaining 26.34% contributed by other species (each<5%). The coefficients of variation for the total, above-, and below-ground biomasses were 37.89%, 37.75%, 41.27%, respectively. The structural ratios of the semivariance functions for total, above-and below-ground biomasses were 0.50, 0.61, and 0.50, respectively, whereas the corresponding ranges were 276, 198, and 375 m. These data indicated a moderately significant spatial autocorrelation for the biomasses, whereas that for the RSR was week. The geostatistical mean of the total biomass density within the fetch of the flux tower (153.63 Mg/hm²) was close to the simple mean of the 106 plots, indicating the effective spatial representation of these plots. The biomass and RSR between the hardwood and mixed deciduous stands did not differ significantly (P>0.1). However, when the basal area (BA) was used as the covariate, the difference in biomass between the two forests was highly significant (P<0.001). The total and above-ground biomasses were more closely correlated with the BA (R² > 85%) than with dominant tree height (R²<41%) for both forest types. The biomass and its components were positively correlated with the slope of the terrain for the mixed deciduous (P<0.001), but was not the hardwood stand. Stepwise regressions for all data of the 106 plots indicated that BA and the dominant tree height were the first and second contributors to the variations in the total and above-ground biomasses, whereas BA and aspect were the most important in the below-ground biomass and RSR. In conclusion, the spatial patterns of biomass and its components were significant within the fetch of the Maoershan flux tower, highlighting the importance of considering spatial variation in the estimation of carbon fluxes using the eddy covariance and biometric methods. The spatial variations were significantly associated with stand basal area, dominant tree height, slope, and aspect, indicating the predictability of the variations.