The isotopic composition of atmospheric nitrous oxide (N₂O) was measured semicontinuously, at ∼35 min frequency in intermittent periods of 1–6 days over one and a half years, using preconcentration coupled to a quantum cascade laser spectrometer at the suburban site of Dübendorf, Switzerland. The achieved measurement repeatability was 0.08‰, 0.11‰, and 0.10‰ for δ¹⁸O, site preference, and δ¹⁵N^bulk respectively, which is better than or equal to standard flask sampling-based isotope ratio mass spectrometry performance. The observed mean diurnal cycle reflected the buildup of N₂O from isotopically light sources on an isotopically heavy tropospheric background. The measurements were used to determine the source isotopic composition, which varied significantly compared to chemical and meteorological parameters monitored at the site. FLEXPART-COSMO transport modeling in combination with modified Emissions Database for Global Atmospheric Research inventory emissions was used to model N₂O mole fractions at the site. Additionally, isotopic signatures were estimated for different source categories using literature data and used to simulate N₂O isotopic composition over the measurement period. The model was able to capture variability in N₂O mole fraction well, but simulations of isotopic composition showed little agreement with observations. In particular, measured source isotopic composition exhibited one magnitude larger variability than simulated, clearly indicating that the range of isotopic source signatures estimated from literature significantly underestimates true variability of source signatures. Source δ¹⁸O signature was found to be the most sensitive tracer for urban/industry versus agricultural N₂O. δ¹⁵N^bulk and site preference may provide more insight into microbial and chemical emission processes than partitioning of anthropogenic source categories.