This study investigates the yield of reactive oxygen species (ROS) under proton FLASH irradiation in well-oxygenated healthy tissue to identify the beam time structure for the lowest ROS yield. Examining the influence of proton beam time structure on radiolytic oxygen consumption and ROS production directly addresses the oxygen depletion hypothesis for the FLASH effect. A bespoke Monte Carlo Geant4-DNA model was used to simulate water radiolysis under a proton FLASH dose rate and output time-resolved G values for the principal radiolysis products. A 1.6 μm cube of water with fixed oxygen concentration was modelled for pulse lengths and periods ranging from 1 ns to 1 ms and 10 ns to 10 ms, respectively. The study was limited to two ROS species (hydroxyl radical [OH^•] and hydrogen peroxide [H₂O₂]) and a pulse-structure grid for which the pulse period was constrained to be 10 times the pulse length. The maximum and total G values, G_max and G_tot, were used to determine the beam time structure that produced the fewest OH^• and H₂O₂. The model shows that the pulse period affected the yield of ROS more than the pulse length. This indicates that the yield of the investigated ROS decreases most under short pulse periods combined with the longest possible pulse duration within those periods. The method of high dose rate proton delivery that is closest to these parameters is an isochronous cyclotron emitting 2 ns pulses separated by 13.7 ns. However, this study also suggests that the oxygen depletion hypothesis cannot fully explain the FLASH effect, as the yield of OH^• in healthy tissue is higher for FLASH irradiation than for conventional irradiation. These conclusions should be interpreted within the limitations of the model, which assumes static oxygenation in a single microscopic volume and explores a restricted set of pulse structures and ROS species. Further work is therefore required to assess the generality of these findings under more realistic biological conditions.