AbstractO2, H2, and H2O2 radiolysis from water ice is pervasive on icy astrophysical bodies, but the lack of a self‐consistent, quantitative model of the yields of these water products versus irradiation projectile species and energy has been an obstacle to estimating the radiolytic oxidant sources to the surfaces and exospheres of these objects. A major challenge is the wide variation of O2 radiolysis yields between laboratory experiments, ranging over 4 orders of magnitude from 5 × 10−7 to 5 × 10−3 molecules/eV for different particles and energies. We revisit decades of laboratory data to solve this long‐standing puzzle, finding an inverse projectile range dependence in the O2 yields, due to preferential O2 formation from an ~30 Å thick oxygenated surface layer. Highly penetrating projectile ions and electrons with ranges ≳30 Å are therefore less efficient at producing O2 than slow/heavy ions and low‐energy electrons (≲ 400 eV) which deposit most energy near the surface. Unlike O2, the H2O2 yields from penetrating projectiles fall within a comparatively narrow range of (0.1–6) × 10−3 molecules/eV and do not depend on range, suggesting that H2O2 forms deep in the ice uniformly along the projectile track, e.g., by reactions of OH radicals. We develop an analytical model for O2, H2, and H2O2 yields from pure water ice for electrons and singly charged ions of any mass and energy and apply the model to estimate possible O2 source rates on several icy satellites. The yields are upper limits for icy bodies on which surface impurities may be present.