Ground-level concentrations of nine volatile organic compounds (VOCs) were measured using a proton transfer reaction-mass spectrometer (PTR-MS) in central London at an urban background (North Kensington, NK, during 16th–25th Jan 2012) and a kerbside site (Marylebone Rd, MRd, during 25th Jan–7th Feb 2012) as part of the winter intensive observation period of the ClearfLo project. Site comparisons indicated that VOC concentrations at the urban background site were significantly lower than at the kerbside site (ratio MRd/NK of 2.3). At the kerbside site PTR-MS measurements of aromatics (benzene, toluene, C2- and C3-benzenes) were compared with the gas chromatography – flame ionization detector data from the UK Government's Automatic Hydrocarbon Network. Very good qualitative agreement was observed between the two methods (r = 0.90–0.91, p < 0.001, N = 260), although there was a significant offset between the instruments. This was partly due to issues with humidity dependent background measurements, but possibly also from isobaric interference of other compounds and their fragments, giving a positive bias to the PTR-MS data. Most compounds showed strong indications of traffic-related sources with double rush hour peaks in diurnal profiles and high correlations with known traffic-related compounds (r = 0.68–0.97 at NK, 0.48–0.87 at MRd, p < 0.001, NNK = 2202–2227, NMRd = 2705–2720) and CO (r = 0.80–0.96 at NK, 0.65–0.84 at MRd, p < 0.001, NNK = 223, NMRd = 256–274). Polar plots agreed with statistical analysis of wind direction dependency and identification of potential emission sources was attempted.