Application of inorganic nitrogen (N) fertilizer and manure can increase nitrous oxide (N2O) emissions. We tested the hypothesis that increased N2O flux from soils amended with manure reflects a change in bacterial community structure and, specifically, an increase in the number of denitrifiers. To test this hypothesis, a field experiment was conducted in a drip-irrigated cotton field in an arid region of northwestern China. Treatments included plots that were not amended (Control), and plots amended with urea (Urea), animal manure (Manure) and a 50/50 mix of urea and manure (U+M). Manure was broadcast-incorporated into the soil before seeding while urea was split-applied with drip irrigation (fertigation) over the growing season. The addition treatments did not, as assessed by nextgen sequencing of PCR-amplicons generated from rRNA genes in soil, affect the alpha diversity of bacterial communities but did change the beta diversity. Compared to the Control, the addition of manure (U+M and Manure) significantly increased the abundance of genes associated with nitrate reduction (narG) and denitrfication (nirK and nosZ). Manure addition (U+M and Manure) did not affect the nitrifying enzyme activity (NEA) of soil but resulted in 39–59 times greater denitrifying enzyme activity (DEA). In contrast, urea application had no impact on the abundances of nitrifier and denitrifier genes, DEA and NEA; likely due to a limitation of C availability. DEA was highly correlated (r = 0.70–0.84, P < 0.01) with the abundance of genes narG, nirK and nosZ. An increase in the abundance of these functional genes was further correlated with soil NO3−, dissolved organic carbon, total C, and total N concentrations, and soil C:N ratio. These results demonstrated a positive relationship between the abundances of denitrifying functional genes (narG, nirK and nosZ) and denitrification potential, suggesting that manure application increased N2O emission by increasing denitrification and the population of bacteria that mediated that process.