Abstract. The Tropospheric ORganic CHemistry experiment (TORCH) took place during the heatwave of summer 2003 at Writtle College, a site 2 miles west of Chelmsford in Essex and 25 miles north east of London. The experiment was one of the most highly instrumented to date. A combination of a large number of days of simultaneous, collocated measurements, a consequent wealth of model constraints and a highly detailed chemical mechanism, allowed the atmospheric chemistry of this site to be studied in detail. Between 25 July and 31 August, the concentrations of the hydroxyl radical and the hydroperoxy radical were measured using laser-induced fluorescence at low pressure and the sum of peroxy radicals was measured using the peroxy radical chemical amplifier technique. The concentrations of the radical species were predicted using a zero-dimensional box model based on the Master Chemical Mechanism version 3.1, which was constrained with the observed concentrations of relatively long-lived species. The model included a detailed parameterisation to account for heterogeneous loss of hydroperoxy radicals onto aerosol particles. Quantile-quantile plots were used to assess the model performance in respect of the measured radical concentrations. On average, measured hydroxyl radical concentrations were over-predicted by 24%. Modelled and measured hydroperoxy radical concentrations agreed very well, with the model over-predicting on average by only 7%. The sum of peroxy radicals was under-predicted when compared with the respective measurements by 22%. Initiation via OH was dominated by the reactions of excited oxygen atoms with water, nitrous acid photolysis and the ozone reaction with alkene species. Photolysis of aldehyde species was the main route for initiation via HO2 and RO2. Termination, under all conditions, primarily involved reactions with NOx for OH and heterogeneous chemistry on aerosol surfaces for HO2. The OH chain length varied between 2 and 8 cycles, the longer chain lengths occurring before and after the most polluted part of the campaign. Peak local ozone production of 17 ppb hr−1 occurred on 3 and 5 August, signifying the importance of local chemical processes to ozone production on these days. On the whole, agreement between model and measured radicals is good, giving confidence that our understanding of atmospheres influenced by nearby urban sources is adequate.