The kinetics and mechanism of the reactions C6H5CH2O2 + C6H5CH2O2 --> 2C6H5CH2O + O2 (3a), C6H5CH2O2 + C6H5CH2O2 --> C6H5CHO + C6H5CH2OH + O2 (3b), and C6H5CH2O2 + HO2 --> C6H5CH2OOH + O2 (4) have been investigated using two complementary techniques : flash photolysis/UV absorption for kinetic measurements and continuous photolysis/FTIR spectroscopy for end-product analyses and branching ratio determinations. The reaction of chlorine atoms with toluene was found to yield benzyl radicals exclusively and was used to generate benzylperoxy radicals in excess oxygen. During this study, relative reaction rate constants of chlorine atoms with compounds related to those involved in the reaction mechanism have been measured at room temperature : k(Cl+toluene) = (6.1 +/- 0.2) X 10(-11), k(Cl+benzaldehyde) = (9.6 +/- 0.4) X 10(-11), k(Cl+benzyl chloride) = (9.7 +/- 0.6) X 10(-12), k(Cl+benzyl alcohol) = (9.3 +/- 0.5) X 10(-11), k(Cl+benzene) < 5 X 10(-16), all in units of cm3 molecule-1 s-1. The products identified following the self-reaction 3 were benzaldehyde, benzyl alcohol, and benzyl hydroperoxide. The latter is the product of the reaction Of C6H5CH2O2 with HO2. The yield of products allowed us to determine the branching ratio alpha = k3a/k3 = 0.4. The UV absorption spectrum of the benzylperoxy radical was determined from 220 to 300 nm. It was similar to those of alkylperoxy radicals, with a maximum cross section at 245 nm of 6.8 X 10(-18) cm2 molecule.-1 Kinetic data were obtained from the detailed simulation of experimental decay traces recorded at 250 nm over the temperature range 273-450 K. The resulting rate expressions are k3 = (2.75 +/- 0.15) X 10(-14) exp[(1680 +/- 140) K/T] cm3 molecule-1 s-1 and k4 = (3.75 +/- 0.32) X 10(-13) exp[(980 +/- 230)K/T) cm3 molecule-1 s-1 (errors = 1sigma). The UV absorption traces in the flash-photolysis kinetic study were well accounted for by the identified products in the FTIR study, thus providing good confidence in the results. However, about 20% of the products have remained unidentified. Some uncertainties persist in the reaction mechanism leading us to assign a fairly large uncertainty of about 50% to the rate constants k3 and k4 over the whole temperature range. This work shows that the aromatic substituent does not provide any specificity in the reactivity of peroxy radicals and confirms that large radicals tend to react faster with HO2 than generally assumed in current atmospheric models.