Naphthalene (NAP), as a surrogate of intermediate-volatility organic compounds (IVOCs), has been proposed to be an important precursor of secondary organic aerosol (SOA). However, the relative contribution of its emission sources is still not explicit. This study firstly conducted the source apportionment of atmospheric NAP using a triple-isotope (δ13C, δ2H, and Δ14C) technique combined with a Bayesian model in the Beijing-Tianjin-Hebei (BTH) region of China. At the urban sites, stable carbon (-27.7 ± 0.7‰, δ13C) and radiocarbon (-944.0 ± 20.4‰, Δ14C) isotope compositions of NAP did not exhibit significant seasonal variation, but the deuterium system showed a relatively more 2H depleted signature in winter (-86.7 ± 8.9‰, δ2H) in comparison to that in summer (-56.4 ± 3.9‰, δ2H). Radiocarbon signatures indicated that 95.1 ± 1.8% of NAP was emitted from fossil sources in these cities. The Bayesian model results indicated that the emission source compositions in the BTH urban sites had a similar pattern. The contribution of liquid fossil combustion was highest (46.7 ± 2.6%), followed by coal high-temperature combustion (26.8 ± 7.1%), coal low-temperature combustion (18.9 ± 6.4%), and biomass burning (7.6 ± 3.1%). At the suburban site, the contribution of coal low-temperature combustion could reach 70.1 ± 6.4%. The triple-isotope based approach provides a top-down constraint on the sources of atmospheric NAP and could be further applied to other IVOCs in the ambient atmosphere.