Stream ecosystems exhibit high degrees of spatial heterogeneity in environmental conditions and biological assemblages at nested scales from watersheds to microhabitats. This heterogeneity may facilitate the co-occurrence of biogeochemical processes that are favored under incompatible environmental conditions, like dinitrogen (N2 gas) fixation and denitrification. We hypothesized that environmental variation at the patch scale (1 - 10’s m) would facilitate the co-occurrence of N2 fixation and denitrification through the formation of hot spots, which we identified as patches with high reaction rates relative to the surrounding spatial matrix. We measured rates of N2 fixation and denitrification and relative abundances of the genes nifH and nirS in patches determined by channel geomorphic units and substrate type in 4 Idaho and 3 Michigan streams encompassing a gradient of N and P concentrations. We found that hot spots of N2 fixation and denitrification where process rates were 1 to 4 times higher than reach-average rates occurred in all 7 study streams. All N2 fixation hot spots were in patches with rock substrates, while denitrification rates and relative abundances of both nifH and nirS were higher in fine sediment patches. Yet, in three of the Idaho streams, rates in the top 25% of all patches for both denitrification and N2 fixation occurred in the same patches, suggesting that variation in conditions at the sub-patch scale can also facilitate co-occurrence of these processes. Predictive modeling across all streams and patches showed that organic matter and dissolved oxygen concentrations were important predictors of rates of N2 fixation, denitrification, and nifH relative abundance, while P concentration was important to N2 fixation, denitrification and nirS. Ammonium concentration was also important to relative abundances of nifH and nirS. Together, our results demonstrate that understanding the spatial ecology of microbially-driven nutrient cycling is required to characterize nutrient fluxes more completely in stream ecosystems.