© 2018 Elsevier Ltd 18O/16O and 17O/16O ratios of atmospheric and dissolved oceanic O2 are used as biogeochemical tracers of photosynthesis and respiration. Critical to this approach is a quantitative understanding of the isotopic fractionations associated with production, consumption, and transport of O2 in the ocean both at the surface and at depth. We made measurements of isotopic fractionations associated with O2 respiration by E. coli. Our study included wild-type strains and mutants with only a single respiratory O2 reductase in their electron transport chains (either a heme-copper oxygen reductase or a bd oxygen reductase). We tested two common assumptions made in interpretations of O2 isotope variations and in isotope-enabled models of the O2 cycle: (i) laboratory-measured respiratory 18O/16O isotopic fractionation factors (18α) of microorganisms are independent of environmental and experimental conditions including temperature, carbon source, and growth rate; And (ii) the respiratory ‘mass law’ exponent, θ, between 18O/16O and 17O/16O, 17α = (18α)θ, is universal for aerobic respiration. Results demonstrated that experimental temperatures have an effect on both 18α and θ for aerobic respiration. Specifically, lowering temperatures from 37 to 15 °C decreased the absolute magnitude of 18α by 0.0025 (2.5‰), and caused the mass law slope to decrease by 0.005. We propose a possible biochemical basis for these variations using a model of O2 reduction that incorporates two isotopically discriminating steps: the reversible binding and unbinding of O2 to a terminal reductase, and the irreversible reduction of that O2 to water. Finally, we cast our results in a one-dimensional isopycnal reaction-advection-diffusion model, which demonstrates that enigmatic δ18O and Δ17O variations of dissolved O2 in the dark ocean can be understood by invoking the observed temperature dependence of these isotope effects.