Results of high-precision measurements of gas production in the BL reaction are presented, and an efficient kinetic model for their analysis is proposed. Based on this model, the data have been examined pulse by pulse, and for the first time, the entire records of gas production could be successfully reduced to series of just a few key parameters. It has been confirmed that the kinetics of O-2(g) production is of the first order with respect to its precursor. Overall, only two steps have been found necessary to fit the observed pulses in gas production. The first step produces the precursor of the recorded O-2(g), and its rate has two components. One component provides the peaks, and its approximation in the form of Gaussian functions has been found as satisfactory. The other component provides the constant baseline of gas production between the pulses. Finally, the precursor gives rise to O-2(g) in the second step, and the simple first-order kinetics suggests that the precursor is otherwise relatively unreactive, making O-2(aq) a logical candidate. However, the rate constant of this process showed almost perfect linearity with the actual concentrations of H2O2, and it was affected only little by variations in the rate of stirring. It thus seems possible that this final step in gas production, responsible for the majority of O-2 produced in pulses, might not be the interphase transport O-2(aq) -> O-2(g), as expected. Instead, it might be a truly chemical process, giving rise to O-2(g) in a reaction of H2O2 with another precursor, which is not involved significantly in any other process, but it is not O-2(aq). If this is true, the second-order rate constant of this process in the system with initial composition of 0.360 M KIO3, 0.345 M H2O2, and 0.055 M HClO4 at 60 degrees C would be 0.25-0.30 M-1.s(-1), depending on the rate of stirring.