New δ 15 N analyses combined with a literature compilation reveal that shale kerogen, VMS-micas, and late-metamorphic vein micas show a secular trend from enriched values in the Archean, through intermediate values in Proterozoic terranes, to the Phanerozoic mode of 3‰–4‰. Kerogen in metashales from the 2.7 Ga Sandur Greenstone Belt, eastern Dharwar Craton, India, is characterized by δ 15 N 13.1‰ ± 1.3‰, and C/N 303 ± 93. A second population has δ 15 N 3.5‰ ± 0.9‰, and C/N 8 ± 0.4, close to the Redfield ratio of modern microorganisms, and is interpreted as precipitates of Proterozoic or Phanerozoic oilfield brines that penetrated the Archean basement. Kerogen from 1.7 Ga carbonaceous shales of the Cuddapah Basin average 5.0‰ ± 1.2‰, close to the mode at 3‰–4‰ for kerogen and bulk rock of Phanerozoic sediments. Biotites from late-metamorphic quartz-vein systems of the 2.6 Ga Kolar gold province, E. Dharwar Craton, that proxy for average crust, are also enriched at 14‰–21‰ for three samples, confirming that the N–budget of the hydrothermal fluids is dominated by sedimentary rocks. Muscovites from altered volcanic rocks in 2.7 Ga Abitibi belt VMS deposits have δ 15 N 12‰–20‰, in keeping with published data for shale kerogen from the same terrane, whereas equivalents in the 1.8 Ga Jerome VMS span 11.7‰–14.1‰. 15 N-enriched values in Precambrian rocks cannot be caused by N-isotopic shifts due to metamorphism or Rayleigh fractionation because (1) pre-, and post-metamorphic samples from the same terrane are both enriched in 15 N; (2) there is no covariation of δ 15 N with N, C/N ratios, or metamorphic grade; and (3) the magnitude of fractionations of 1‰ (greenschist) to 3‰ (amphibolite facies) during progressive metamorphism of sedimentary rocks, as constrained from empirical observations and experimental studies, is very small. Nor can 15 N-enriched values stem from long-term preferential diffusional loss of 14 N, as samples were selected from terranes where 40 Ar/ 39 Ar ages are within a few million years of concordant U-Pb ages; nitrogen is structurally bound in micas, whereas Ar is not. It is possible that the 15 N-enriched values stem from a different N-cycle in the Archean, with large biologically mediated fractionations, yet the magnitude of the fractionations between atmospheric N 2 and organic nitrogen observed exceeds any presently known, and chemoautotrophic communities tend to depleted values. Earlier results on Archean cherts show a range of δ 15 N from −6‰ to 30‰. Given the temporal association of chert–banded iron formation (BIF) with mantle plumes, the range is consistent with mixing between mantle N 2 of −5‰ and the 15 N-enriched marine reservoir identified in this study. The 15 N-enriched Archean atmosphere-hydrosphere reservoir does not robustly constrain Archean redox-state. We attribute the 15 N-enriched reservoir to a secondary atmosphere derived from CI-chondrite-like material and comets with δ 15 N of +30‰ to +42‰. Shifts of δ 15 N to its present atmospheric value of 0‰ can be accounted for by a combination of early growth of the continents with sequestration of atmospheric N 2 into crustal rocks, and degassing of mantle N ∼−5‰. If Earth's surface environment became oxygenated ca. 2 Ga, then there were no associated large N-isotope excursions.