The use of stable isotopic tracer methods is becoming a popular approach for in situ evapotranspiration (ET) partitioning studies at the ecosystem scale. Ecosystem evapotranspiration is usually partitioned based on different isotopic composition among the aggregated ET flux (δET) and its two components: physical evaporation process (δE) and biological transpiration process (δT). Uncertainties in calculating these three isotopic compositions, especially δET, are usually not negligible and could propagated to large uncertainty in the ET partitioning outcomes. In this study, we present a new ET partitioning approach utilizing dual isotope-based evapotranspiration partitioning based on d-excess (d-excess = δ2H - 8 × δ18O) . A field deployable laser absorption spectrometer was used for in situ measurements of isotopic composition (δ2H and δ18O) of atmospheric water vapor at different heights within the turbulently mixing ecosystem boundary layer for tallgrass prairie. This study incorporated a new refinement to the Craig-Gordon model to describe δE. During non-growing season, estimates of δET—based on Keeling plot and flux-gradient approaches—were compared against δE values derived by the Craig-Gordon model, under the assumption of no transpiration and thus the equality between δET and δE. During the growing season, coupled with isotopic sampling in plant and soils, we partitioned ET into transpiration and evaporation. This refined dual isotope-based approach of ET partitioning shows promise in reducing the uncertainties of the fractional contributions of evaporation and transpiration, thus enhance our understanding on the mechanisms underlying plant water use efficiency and the role of vegetation ecophysiological processes in eco-hydrologic processes under the changing environment.