Concentrations and fluxes of seven volatile organic compounds (VOCs) were measured between August and December 2012 at a roof-top site in central London as part of the ClearfLo project (Clean Air for London). VOC concentrations were quantified using a proton transfer reaction-mass spectrometer and fluxes were calculated using a virtual disjunct eddy covariance technique. The median VOC fluxes, including aromatics, oxygenated compounds and isoprene, ranged from 0.07 to 0.33 mg m−2 h−1 and mixing ratios were 7.27 ppb for methanol (m / z 33) and <1 ppb for the remaining compounds. Strong relationships were observed between most VOC fluxes and concentrations with traffic density, but also with photosynthetically active radiation (PAR) and temperature for the oxygenated compounds and isoprene. An estimated 50–90 % of aromatic fluxes were attributable to traffic activity, which showed little seasonal variation, suggesting boundary layer effects or possibly advected pollution may be the primary causes of increased concentrations of aromatics in winter. PAR and temperature-dependent processes accounted for the majority of isoprene, methanol and acetaldehyde fluxes and concentrations in August and September, when fluxes and concentrations were largest. Modelled biogenic isoprene fluxes using the G95 algorithm agreed well with measured fluxes in August and September, due to urban vegetation. Comparisons of estimated annual benzene emissions from the London and National Atmospheric Emissions Inventory agreed well with measured benzene fluxes. Flux footprint analysis indicated emission sources were localized and that boundary layer dynamics and source strengths were responsible for temporal and spatial VOC flux and concentration variability during the measurement period.