Nitrogen (N) losses constrain rates of plant carbon dioxide (CO2) uptake and storage in many ecosystems globally. N isotope models have been used to infer that ~30 % of terrestrial N losses occur via microbial denitrification; however, this approach assumes a small isotope effect associated with N leaching losses. Past work across tropical/sub-tropical forest sites has confirmed this expectation; however, the stable N isotope ratio (δ15N) of ecosystem leaching has yet to be systematically evaluated in seasonally dry climates or across major ecosystem disturbances. We here present new measurements of the δ15N of total dissolved N (TDN) in small streams, bulk deposition, and soil pools across eight watershed sites in California, including grassland, chaparral, and coastal redwood forest ecosystems, with and without fire, grazing, and forest harvesting. Regardless of the dominant vegetation type or disturbance regime, average δ15N of TDN in stream water differed only slightly (<~1 ‰) from that of bulk soil δ15N, revealing a uniformly small isotope effect associated with N leaching losses even under non-steady state conditions. Rather, lower input δ15N compared to TDN δ15N in streams pointed to fractionations via gaseous loss pathways as the dominant mechanism behind soil δ15N enrichment. We conclude that N leaching does not impart a major isotope effect across a broad range of ecosystems and conditions examined, thereby advancing the N gas-loss hypothesis as the principal explanation for variation in bulk soil δ15N.