Reactive nitrogen species have a strong influence on atmospheric chemistry and climate, tightly coupling the Earth's nitrogen cycle with microbial activity in the biosphere. Their sources, however, are not well constrained, especially in dryland regions accounting for a major fraction of the global land surface. Here, we show that biological soil crusts (biocrusts) are emitters of nitric oxide (NO) and nitrous acid (HONO). Largest fluxes are obtained by dark cyanobacteria-dominated biocrusts, being ∼20 times higher than those of neighboring uncrusted soils. Based on laboratory, field, and satellite measurement data, we obtain a best estimate of ∼1.7 Tg per year for the global emission of reactive nitrogen from biocrusts (1.1 Tg a(-1) of NO-N and 0.6 Tg a(-1) of HONO-N), corresponding to ∼20% of global nitrogen oxide emissions from soils under natural vegetation. On continental scales, emissions are highest in Africa and South America and lowest in Europe. Our results suggest that dryland emissions of reactive nitrogen are largely driven by biocrusts rather than the underlying soil. They help to explain enigmatic discrepancies between measurement and modeling approaches of global reactive nitrogen emissions. As the emissions of biocrusts strongly depend on precipitation events, climate change affecting the distribution and frequency of precipitation may have a strong impact on terrestrial emissions of reactive nitrogen and related climate feedback effects. Because biocrusts also account for a large fraction of global terrestrial biological nitrogen fixation, their impacts should be further quantified and included in regional and global models of air chemistry, biogeochemistry, and climate.