One-third of global lake nitrous oxide emissions are concentrated in the northern temperate zone (40°N-50°N).

Lacustrine nitrous oxide (N2O) emissions remain highly uncertain in the global greenhouse gas budget largely because of sparse observations and the omission of small lakes. Here we compile the Global Lake N2O database (GLON2O), integrating 6550 in situ measurements from 3829 lakes worldwide, and develop a machine learning framework to estimate emissions from lakes ≥0.1 square kilometers. To quantify previously overlooked systems, we construct the Global Small Lakes database (GSLAKES). We estimate that global lakes emit 0.208 teragrams of nitrogen per year of N2O, with approximately one-third originating from the northern temperate zone (40° to 50°N). Despite accounting for only 9.6% of total lake area, small lakes (<0.1 square kilometers) contribute 34.6% of global emissions, revealing a pronounced U-shaped relationship between emission intensity and lake size. We further identify hierarchical predictors of lake N2O emissions: Natural constraints determine biogeochemical potential, anthropogenic nitrogen inputs amplify emissions, and lake morphometry regulates spatial expression. These findings refine global lake N2O inventories and provide an empirical basis for scale-aware mitigation.