Abstract. Aerosol source apportionment remains a critical challenge for understanding the transport and aging of aerosols, as well as for developing successful air pollution mitigation strategies. The contributions of fossil and non-fossil sources to organic carbon (OC) and elemental carbon (EC) in carbonaceous aerosols can be quantified by measuring the radiocarbon (14C) content of each carbon fraction. However, the use of 14C in studying OC and EC has been limited by technical challenges related to the physical separation of the two fractions and small sample sizes. There is no common procedure for OC/EC 14C analysis, and uncertainty studies have largely focused on the precision of yields. Here, we quantified the uncertainty in 14C measurement of aerosols associated with the isolation and analysis of each carbon fraction with the Swiss_4S thermal–optical analysis (TOA) protocol. We used an OC/EC analyzer (Sunset Laboratory Inc., OR, USA) coupled to a vacuum line to separate the two components. Each fraction was thermally desorbed and converted to carbon dioxide (CO2) in pure oxygen (O2). On average, 91 % of the evolving CO2 was then cryogenically trapped on the vacuum line, reduced to filamentous graphite, and measured for its 14C content via accelerator mass spectrometry (AMS). To test the accuracy of our setup, we quantified the total amount of extraneous carbon introduced during the TOA sample processing and graphitization as the sum of modern and fossil (14C-depleted) carbon introduced during the analysis of fossil reference materials (adipic acid for OC and coal for EC) and contemporary standards (oxalic acid for OC and rice char for EC) as a function of sample size. We further tested our methodology by analyzing five ambient airborne particulate matter (PM2.5) samples with a range of OC and EC concentrations and 14C contents in an interlaboratory comparison. The total modern and fossil carbon blanks of our setup were 0.8 ± 0.4 and 0.67 ± 0.34 μg C, respectively, based on multiple measurements of ultra-small samples. The extraction procedure (Swiss_4S protocol and cryo-trapping only) contributed 0.37 ± 0.18 μg of modern carbon and 0.13 ± 0.07 μg of fossil carbon to the total blank of our system, with consistent estimates obtained for the two laboratories. There was no difference in the background correction between the OC and EC fractions. Our setup allowed us to efficiently isolate and trap each carbon fraction with the Swiss_4S protocol and to perform 14C analysis of ultra-small OC and EC samples with high accuracy and low 14C blanks.