“FLASH radiotherapy” is a new method of radiation treatment by which large doses of radiation are delivered at high dose rates to tumors almost instantaneously (a few milliseconds), paradoxically sparing healthy tissue while preserving anti-tumor activity. To date, no definitive mechanism has been proposed to explain the different responses of the tumor and normal tissue to radiation. As a first step, and given that living cells and tissues consist mainly of water, we studied the effects of high dose rates on the transient yields (G values) of the radical and molecular species formed in the radiolysis of deaerated/aerated water by irradiating protons, using Monte Carlo simulations. Our simulation model consisted of two steps: 1. The random irradiation of a right circular cylindrical volume of water, embedded in nonirradiated bulk water, with single and instantaneous pulses of N 300-MeV incident protons (“linear energy transfer” or LET ∼ 0.3 keV/µm) traveling along the axis of the cylinder; and 2. The development of these N proton tracks, which were initially contained in the irradiated cylinder, throughout the solution over time. The effect of dose rate was studied by varying N, which was calibrated in terms of dose rate. For this, experimental data on the yield G(Fe³⁺) of the super-Fricke dosimeter as a function of dose rate up to ∼10¹⁰ Gy/s were used. Confirming previous experimental and theoretical studies, significant changes in product yields were found to occur with increasing dose rate, with lower radical and higher molecular yields, which result from an increase in the radical density in the bulk of the solution. Using the kinetics of the decay of hydrated electrons, a critical time (τ_c), which corresponds to the “onset” of dose-rate effects, was determined for each value of N. For the cylindrical irradiation model, τ_c was inversely proportional to the dose rate. Moreover, the comparison with experiments with pulsed electrons underlined the importance of the geometry of the irradiation volume for the estimation of τ_c. Finally, in the case of aerated water radiolysis, we calculated the yield of oxygen consumption and estimated the corresponding concentration of consumed (depleted) oxygen as a function of time and dose rate. It was shown that this concentration increases substantially with increasing dose rate in the time window ∼1 ns–10 µs, with a very pronounced maximum around 0.2 µs. For high-dose-rate irradiations (>10⁹ Gy/s), a large part of the available oxygen (∼0.25 mM for an air-saturated solution) was found to be consumed. This result, which was obtained on a purely water radiation chemistry basis, strongly supports the hypothesis that the normal tissue-sparing effect of FLASH stems from temporary hypoxia due to oxygen depletion induced by high-dose-rate irradiation.