Nitrous oxide (N_2O) is a major greenhouse gas, which is not only an important contributor to global warming, but also a damager of the stratospheric ozone layer. Upland is an important source of N_2O, of which the emission activity has been widely reported to be related to soil structure, and denitrification is a major pathway of N_2O production in upland soil. Soil aggregates are an important component in soil structure and a factor affecting soil N_2O emission. However, little has been reported on mechanisms of soil microbes explaining differences in N_2O emission rate between different fractions of soil aggregate. In this study, soil aggregates, <1 mm, 2 ~4 mm, and 4 ~8 mm in diameter, separated from the soil of a vegetable field, over 15 years in cultivation history, were sprayed with NH_4NO_3 solution to add N 700 mu g g~(-1) soil and regulate soil water content up to 35%. Afterwards, 50 g of aggregates on a dry weight basis was taken from each fraction and transferred into a 500 ml jar for soil incubation under 30 ℃ for a 96 h, and each had 3 replicates. Gas samples were collected once every 12 h after during the incubation and N_2O concentrations in the samples were determined with a Gas-chromatograph. Soil samples for molecular analysis were collected once every 24 h and quick-frozen in liquid nitrogen, and stored at -8O ℃. Quantitative Polymerase Chain Reaction (qPCR) and Terminal-Restricted Fragment Length Polymorphism (T-RFLP) were used to determine changes in abundance and composition of denitrifying genes, narG and nosZ. Results show that N_2O emission rate was closely related to abundance of narG- and nosZ-containing soil microbes, while the fraction of <1 mm aggregates was significantly higher in N_2O emission rate than the fractions of 2 ~ 4 mm and 4 ~ 8 mm, and the fraction of 4 ~8 mm was the lowest. N_2O emission peaked at 12 h and 24 h of incubation. At 12 h, N_2O emission rate of <1mm aggregates reached 7.637 mu g g~(-1) h~(-1), about 36% higher than that of 2 ~4 mm aggregates, while the N_2O emission rate of 2 ~ 4 mm aggregates was about 2 times higher than that of 4 ~8 mm aggregates, reaching 1.965 mu g g~(-1) h~(-1). The differences between the fractions of aggregates were gradually narrowed with the incubation going on, and to the least after 96 h of incubation. Interestingly, the relationships between abundances of narG and particle size of soil aggregates were very similar to those between N_2O emission rates and particle size of soil aggregates. The fraction of <1 mm aggregates was the highest in abundance of narG-containing bacteria, much higher than the fraction of 2 ~4 mm aggregates, while the latter was much higher than the fraction of 4 ~8 mm aggregates in all the soil samples. The abundance of nosZ in soil aggregates demonstrated a similar trend that the fraction smallest in particle size was the highest in copy numbers of nosZ after 48 h of incubation, while the fraction the largest in particle size the lowest. The 2 ~ 4 mm and 4 ~ 8 mm fractions of aggregates were relatively stable in nosZ abundance, lingering around 2.5 * 10~8 g~(-1) and 1.6 * 108 g~(-1) during the whole incubation process, while the <1 mm fraction of aggregates leveled off around 3.6 * 10~8 g~(-1) after 48 h of incubation. N_2O emission rates of the three fractions of aggregates were significantly and positively related to narG gene abundance and to nosZ gene abundance as well at 48 h, 72 h and 96 h of the incubation. However, T-RFLP shows that at most of the time points during the incubation, the fractions differed in number of strips low in content of narG and nosZ, but did not in number of major strips containing narG and nosZ. When narG and nosZ were used as biomarkers characterizing structure and composition of the soil microbial community, no difference was found in structure of soil microbial community between the fractions of soil aggregates, and moreover, the main tendency won' t change with the incubation going on.