The aim of the present paper is to study radical reactions important in the mechanism of the Belousov-Zhabotinsky (BZ) reaction with its simplest organic substrate, oxalic acid, and to model the oscillatory system applying the newly determined rate constants. We considered five radical species in this BZ system: carboxyl radical, bromine atom, dibromine radical ion, and bromine monoxide and dioxide radicals. To study separately reactions of only three radicals, (CO2H)-C-., Br-., and Br-.(2)-, semibatch experiments were performed. The semibatch reactor contained oxalic acid, elemental bromine, and bromide ions in a solution of 1M H2SO4 at 20 degreesC, and a continuous inflow of Ce4+ generated carboxyl radicals. The carboxyl radicals initiate a chain reaction: first they react with elemental bromine and produce bromine atoms (CR1); then the bromine atoms react with oxalic acid, producing carboxyl radicals again (CR2). Consumption of elemental bromine in the chain reaction was followed with a bright Pt electrode. By measuring the stoichiometry of the chain reaction, it was possible to determine or estimate several rate constants. It was found that CR1 is a fast reaction with an estimated k value of more than 10(9) M-1 s(-1). The rate constant of CR2 is 7 x 10(5) M-1 s(-1), and the k value for the Ce4+-(CO2H)-C-. reaction is 1.5 x 10(9). These values were obtained by comparing experiments with model calculations. Such simulations also suggested that a reaction of Br-.(2)- with oxalic acid, analogous to CR2, plays a negligible role or no role here. Simulations of the oscillatory system applied rate constants, which were known from the literature, or determined here or in the first part of our work, and some unknown rate constants were estimated on the basis of analogous radical reactions. To obtain an optimal fit between experiments and simulations, only one rate constant was used as a variable parameter. This was the reaction of carboxyl radical with acidic bromate with an optimal k value of 1.0 x 10(7) M-1 s(-1). Agreement between experimental and simulated oscillations was satisfactory at low bromine removal rates (that rate was controlled by a nitrogen gas flow), but a disagreement was found at higher flow rates. Possible reasons for this disagreement are discussed in the conclusion.