Brown carbon (BrC) is a significant component of atmospheric carbonaceous aerosols and has important influence on climate, atmospheric chemistry, and air quality. Our knowledge about its sources and atmospheric evolution is limited due to the challenges of accurate estimation of its mass concentrations. To address this gap, we examined concurrent measurements of bihourly BrC and a suite of organic molecular tracers in urban Shanghai from March to August 2020, BrC was determined by synergistically applying the Bayesian inference (BI) model to multi-wavelength aethalometer and total carbon (TC) measurements. BrC constituted 38–40 % of TC, with pronounced seasonal differences in diurnal evolution. The source apportionment analysis resolved 15 source factors and among them 7 are secondary sources, namely, secondary nitrate and sulfate formation processes, aromatic secondary organic aerosol (SOA), α-pinene SOA, β-caryophyllene SOA, isoprene SOA, oxygenated cooking OA, and secondary biomass burning. Secondary BrC accounted for 42 % and 49 % of total BrC in spring and summer, respectively. Seasonal contrast of the BrC contributions from primary and secondary sources were noticed. The formation mechanisms of secondary BrC were investigated by isolating events with continuous increment in secondary BrC concentrations, and the analysis suggested the diversity of oxidative pathways governing secondary BrC production. This study demonstrates the potential applicability of coupling the BI framework with online measurements of organic tracers for robust, chemically resolved BrC source apportionment. Our field observations provide the first quantitative source analysis and offer new insights into the primary sources and secondary formation of BrC in the urban atmosphere.